Category: AI

SSPAI Morning Brief: Apple Cuts App Store Fees in China
Morning Brief
1. Apple announces lower App Store commission rates in China
2. Samsung adjusts low-level flashing mechanism for Galaxy devices
3. PEGI to revise game age rating rules starting in June
4. NVIDIA open-sources Nemotron 3 Super AI model weights
5. Feishu launches official OpenClaw plugin
6. Tencent’s WorkBuddy (Lobster edition) adds direct WeChat connectivity
7. OpenAI CEO says AI faces multiple obstacles in the United States
8. Microsoft to roll out Xbox FSE feature to all Windows 11 PCs
9. Microsoft announces the return of the Xbox backward compatibility program
10. Steam Machine Verified standards for gaming consoles published
11. Firefox 115 ESR support extended to August 2026
12. Other News Worth a Quick Look
Apple announces lower App Store commission rates in China

Apple announced on its developer website that, after discussions with Chinese regulators, the company has decided to adjust App Store distribution commission rates in mainland China. Specifically:

The standard commission rate for in-app purchases and paid apps will be reduced from 30% to 25%;
At the same time, for developers eligible under the App Store Small Business Program and the Mini Apps Partner Program, the commission rate for in-app purchases and the commission applied to auto-renewing subscriptions starting from the second year will be reduced from 15% to 12%.

Apple stated that the new rates will take effect on March 15, 2026, and developers will not need to re-sign related agreements. The company also emphasized that it will continue to provide fair and transparent terms for developers distributing apps in China, and pledged that overall App Store commission levels in China will not exceed those in other markets. Source

Samsung adjusts low-level flashing mechanism for Galaxy devices

Tech outlet Android Authority reported on March 11 that Samsung has adjusted the low-level flashing mechanism for Galaxy devices in the latest One UI 8.5 firmware. Tools such as Odin—commonly used for flashing official firmware, installing custom ROMs, or recovering devices—and the associated Download Mode entry point were briefly believed to have been disabled. Affected models include the Galaxy Z Fold 7 and the newly released Galaxy S26 series. When users attempt to enter Download Mode, the traditional interface no longer appears; instead, a blank blue screen with an exit prompt is displayed.

Further investigation shows that Download Mode and Odin still exist, but users must first enable the device’s Maintenance Mode before entering the relevant mode. Samsung has not yet released an official explanation for the change. Discussions generally suggest the adjustment may relate to strengthening system security, restricting firmware sideloading, and tightening ecosystem control. The change may also affect developers, repair services, and users who rely on Odin for device recovery or system downgrades. Source

PEGI to revise game age rating rules starting in June

The Pan European Game Information (PEGI) organization announced that it will update its game age rating standards starting in June. Under the new rules, games containing paid random items (such as card packs, gacha systems, or randomized reward keys) will receive a PEGI 16 rating, while social casino-style games will receive a PEGI 18 rating. The updated system will also consider factors such as in-game purchases, online communication features, and design elements that encourage frequent player return—such as daily missions or consecutive login rewards.

For example, games featuring time-limited or quantity-limited purchase mechanics will typically receive a PEGI 12 rating; if spending features are disabled by default, the rating may be lowered to PEGI 7. If in-game purchases are tied to NFT assets that are required for gameplay and are tradable, the game will be rated PEGI 18.

The new rules will apply only to games submitted for classification after June. PEGI said the changes reference recent reforms to the rating system of Germany’s Unterhaltungssoftware Selbstkontrolle (USK). The updated standards will place greater emphasis on online interaction risks and the influence of game design on consumer spending and gameplay behavior, rather than focusing solely on traditional factors such as violence or drug content.

PEGI noted that after Germany updated its rating system, about 30% of games were affected by the new standards, with roughly one-third receiving higher age ratings. PEGI believes the updated rules will provide clearer information for parents to help them determine whether a game is appropriate for children, while also acknowledging that some game developers may question higher age classifications. Source

In addition, Valve recently published a statement on the Steam support website responding to a lawsuit filed earlier by the New York State Attorney General regarding loot box systems in games such as Counter-Strike 2 and DOTA 2. Valve stated that paid loot box mechanics in games are similar to opening baseball card packs, blind boxes, or trading card packs in the real world, and therefore do not constitute gambling. Valve emphasized that players can fully enjoy these games without spending money, since items obtained from loot boxes are purely cosmetic and do not affect gameplay balance. The company also said its data shows that most players never open any loot boxes and simply play the games normally, meaning non-paying players are not at a disadvantage.

Previously, the New York Attorney General filed the lawsuit in February, accusing Valve of “facilitating gambling behavior” through the Steam platform and popular online games, and seeking a permanent ban on promoting such mechanics along with financial penalties. In its statement, Valve said the company has long taken action against third-party gambling websites, banning more than one million Steam accounts involved in gambling, fraud, or theft, and introducing mechanisms such as trade reversal and trade cooldown systems to limit related activities. Source

NVIDIA open-sources Nemotron 3 Super AI model weights

On March 11, NVIDIA published a blog post announcing the release of Nemotron 3 Super, the company’s most powerful open-weight AI model to date.

Nemotron 3 Super features 120 billion parameters and uses a Mixture-of-Experts (MoE) architecture, activating only about 12 billion parameters during inference. The model is designed for scenarios involving large-scale deployment of complex agentic AI systems. Combined with advanced reasoning capabilities, it helps autonomous agents complete tasks with higher accuracy. The model provides a context window of up to 1 million tokens, allowing agents to retain the full state of workflows in memory, avoid goal drift, and significantly reduce the cost of multi-step reasoning.

NVIDIA stated that companies such as Perplexity, Palantir, and Siemens have already adopted the model in real-world applications, covering core workflows in areas including search, software development, telecommunications, and semiconductor design.

NVIDIA has taken a highly open approach with Nemotron 3 Super. The company has released the model weights under a permissive license and published the full training methodology, including a training dataset exceeding 10 trillion tokens as well as evaluation recipes. Developers can currently access the model through Hugging Face and several cloud platforms, including Google Cloud and Oracle Cloud, with support for AWS and Azure coming soon. In addition, the model is available as an NVIDIA NIM microservice, allowing seamless deployment in on-premises data centers or cloud environments. Source

Feishu launches official OpenClaw plugin

Feishu announced the launch of an official OpenClaw plugin that allows OpenClaw to directly access collaboration tools within the Feishu ecosystem, including documents, group chats, calendars, tasks, and multi-dimensional tables. After granting plugin permissions, OpenClaw can read and analyze messages and document content within Feishu, while also performing actions such as writing documents, sending messages, creating schedules, and generating multi-dimensional tables. This reduces the need for users to repeatedly copy and paste content between AI tools and workplace software.

The official plugin also supports message reading and replies, cloud document creation and updates, multi-dimensional table data management, calendar and participant management, and task management. These capabilities allow AI to better understand contextual information within collaborative workflows.

Feishu stated that the goal of the plugin is to expand AI from a simple chat interface into an intelligent assistant capable of participating in real work tasks. However, the company also warned that because the plugin accesses work-related data such as messages, documents, and schedules, potential data security and privacy risks remain. In addition, since AI may produce misjudgments or “hallucinations,” users are advised to preview and confirm actions—such as sending messages or modifying content—before execution, rather than relying on full automation. Feishu also recommends testing the plugin within a personal account environment before gradually integrating it into real workplace settings. Source

Tencent’s WorkBuddy (Lobster edition) adds direct WeChat connectivity

Tencent’s WorkBuddy (Lobster edition) announced a new version early on March 12. The update introduces one-click WeChat connectivity, allowing users to remotely command the system through WeChat. The enterprise WeChat version also gains persistent connection capabilities along with automated task execution and monitoring features.

Among the updates, the “one-click WeChat connection” feature stands out as particularly useful. To enable it, users open Claw Settings in the upper-right corner, select WeChat Customer Service Integration, and scan a QR code with their phone to complete the binding process. Once connected, a dedicated contact appears within WeChat. Tencent states that from that point forward, users can send simple text commands from anywhere, and WorkBuddy on their office computer will immediately execute tasks such as searching for information, conducting research, writing content, or processing files. All operations run locally on the computer. As long as the computer remains powered on, the system can operate continuously 24 hours a day, 7 days a week. Source

OpenAI CEO says AI faces multiple obstacles in the United States

At the BlackRock U.S. Infrastructure Summit held in Washington, Sam Altman said that as AI technology develops rapidly, negative public narratives have begun to emerge. For example, some critics argue that data center construction drives up electricity prices, while certain companies cite AI as a reason when announcing layoffs. The combination of these factors has gradually turned AI into a controversial topic in public discussions. In addition, debates between governments and technology companies over regulation and the boundaries of power remain unresolved, adding further uncertainty to the future of the AI industry.

Multiple opinion polls indicate that the American public is becoming more cautious about AI. A survey by NBC News found that 57% of respondents believe the risks of AI outweigh its potential benefits. Research from the Pew Research Center shows that about half of U.S. adults say they feel “more concerned than excited” about the development of AI, a figure that has increased by 13 percentage points since 2021. Altman also said that the United States still leads in the global AI race, but if AI adoption in industry and the economy does not progress quickly enough, that advantage could weaken. He believes AI still holds strong potential to drive economic growth and may reshape how society operates in the future. Source

Microsoft to roll out Xbox FSE feature to all Windows 11 PCs

On March 12, Microsoft announced that the Xbox FSE full-screen experience (Xbox Full Screen Experience), originally introduced last June for handheld gaming devices, has been officially renamed “Xbox Mode” and will begin rolling out to all Windows 11 PCs starting next month.

According to publicly available information, Microsoft’s FSE full-screen experience is similar to Steam’s Big Picture mode and aims to provide a more user-friendly interface for gaming PCs connected to large displays. The mode supports not only games from Microsoft’s own platform but also titles from Steam, Epic Games Store, GOG, Ubisoft, and Blizzard’s Battle.net.

With this mode, players can switch between different gaming platforms using only a controller, eliminating the need to frequently navigate menus with a mouse or touchscreen. The interface has also been redesigned specifically for controller input. Meanwhile, the system reduces background processes and postpones non-essential tasks to improve overall gaming performance. Source

Microsoft announces the return of the Xbox backward compatibility program

On March 12, Microsoft announced at GDC 2026 that it will restart the Xbox Backwards Compatibility Program. The initiative aims to allow players to access more classic titles from the original Xbox and Xbox 360 on current and future Xbox platforms.

Microsoft has not yet revealed the specific implementation details of the revived program and has only stated that it will return in some form later this year.

Previous reports suggest that the new Xbox backward compatibility plan may support running classic games directly on devices. At the same time, Microsoft’s next-generation Xbox hardware is rumored to be a compact gaming PC running Windows 11 with an Xbox-style interface. Under this architecture, a plausible scenario is that Microsoft could release official emulators for original Xbox and Xbox 360 games on Windows 11, enabling broader backward compatibility support. Source

Steam Machine Verified standards for gaming consoles published

At GDC 2026, Valve unveiled the hardware verification requirements for the Verified label for the Steam Machine gaming console and the Steam Frame VR headset. The Verified label is a game compatibility certification system introduced by Valve for devices such as Steam Deck, Steam Machine, and Steam Frame, indicating that a game can run smoothly on the device without additional configuration, with optimized controls, clear display, and full controller support.

For Steam Machine consoles, Valve has set a performance benchmark of maintaining 30 FPS at a native resolution of 1080p. Valve also confirmed that the certification standards for Steam Deck Verified and Steam Machine Verified are identical. All games that have already received the Steam Deck Verified label will automatically be compatible and will directly receive the Steam Machine Verified label as well. In addition, the certification includes several core hardware requirements, such as overall hardware performance reaching six times that of the Steam Deck and adopting the same controller input standards as the Steam Deck.

For the Steam Frame VR headset, Valve stated that if a game runs through PC streaming, no additional verification is required. However, for standalone operation on the device itself, games must meet certain performance benchmarks. VR games must maintain a stable 90 FPS, while 2D games must reach 30 FPS at a resolution of 720p.

Valve also recommends that developers optimize their games for SteamOS devices, such as enabling offline play after the first launch, improving performance on lower-end hardware, and enhancing support for controller input. Source

Firefox 115 ESR support extended to August 2026

On March 11, Mozilla announced that it will further extend the support cycle for Mozilla Firefox 115 ESR on older operating systems. The extension applies to systems such as Windows 7, Windows 8.1, and macOS 10.12–10.14, with the support period extended from the originally planned March 2026 to August 2026.

Mozilla also emphasized on its support page that the team plans to conduct another evaluation in July 2026 and will announce the final end-of-support schedule afterward. Therefore, August 2026 may not necessarily be the final deadline, as the timeline will continue to be adjusted based on user data and usage patterns. Source

Other News Worth a Quick Look
- Mark Gurman reports that Apple’s upcoming foldable iPhone may adopt an interface layout similar to the iPad, allowing users to view two apps simultaneously to enhance multitasking. The outer display is expected to be close in size to a smaller iPhone, while the inner display may use a wider aspect ratio similar to Google’s first-generation Pixel Fold rather than the proportions used by some existing foldable flagships. The device is not expected to run current iPad apps directly. However, Apple plans to introduce a sidebar design for its “core” apps and will provide developers with related tools so that iPhone apps can deliver an experience closer to iPad apps when displayed on a larger screen. The report also notes that because the device’s front panel will be extremely thin, Apple may not use Face ID. Instead, it may integrate Touch ID into the side power button and adopt a hole-punch front camera design, while still supporting the Dynamic Island feature. The foldable iPhone is expected to feature two rear cameras and may be priced at around $2,000. Source
- DJI has recently begun teasing its first 8K flagship panoramic drone, the DJI Avata 360, and confirmed that the new product will launch on March 26. Official teaser posters indicate that the drone continues the Avata series’ integrated propeller guard design and familiar body structure, while potentially offering improved control and flight experience along with panoramic aerial shooting capabilities. Previous leaks suggest that the DJI Avata 360 uses a gray ducted-frame structure with fully enclosed spherical propeller guards. The gimbal reportedly supports a rotating storage design that protects the camera from scratches caused by gravel during takeoff and landing. Rumors also indicate the drone will feature a 1-inch sensor, support panoramic imaging, record up to 6K 60FPS HDR video, and include omnidirectional obstacle avoidance. Source
March 13, 2026
SSPAI Morning Brief: Google Launches Gemini Embedding 2 as ComfyUI Introduces App-Style Workflows
Morning Brief
1. Google releases Gemini Embedding 2 model
2. MediaTek unveils multiple IoT SoC platforms
3. ComfyUI adds support for app-style workflows
4. RØDE launches Video Core
5. iReader releases iReader Neo 3 Ultra
Google releases Gemini Embedding 2 model

On March 11, Google announced the new Gemini Embedding 2 model. It is Google’s first native multimodal embedding model, capable of mapping text, images, videos, and documents into the same embedding space.

Gemini Embedding 2 supports text, images, video, audio, and documents, and can recognize semantic intent across 100 languages. The new model simplifies complex data processing pipelines while enhancing multimodal application capabilities. Use cases include retrieval-augmented generation (RAG), semantic search, sentiment analysis, and data clustering. Gemini Embedding 2 can also accept multiple input types—such as “image plus text”—within a single request, enabling analysis of relationships across different media formats.

The model is currently available in public preview through the Gemini API and Vertex AI. Source

MediaTek unveils multiple IoT SoC platforms

On March 9, MediaTek unveiled a series of IoT SoC platforms at Embedded World 2026 in Nuremberg, Germany, including the high-end Genio Pro solution as well as the lower-positioned Genio 420 and Genio 360.

Genio Pro is built on TSMC’s advanced 3nm process and features CPU and GPU specifications similar to the mobile Dimensity 9400. It is designed to operate under industrial wide-temperature environments. The platform supports up to 16 cameras and as many as three 4K displays, and is compatible with a range of Linux distributions as well as the open-source ROS robotics operating system.

Genio 420 uses a 6nm process and provides 7.2 TOPS of system AI computing power while integrating 16GB LPDDR5X memory. It remains pin-compatible with the existing Genio 720 and Genio 520, allowing users to migrate between chip models more easily.

The lower-tier Genio 360 and Genio 360P deliver system AI performance of 6 TOPS and 8.5 TOPS respectively. They support up to 8GB LPDDR4X-3733 memory and can meet the requirements for running 2B parameter models on-device. Source

ComfyUI adds support for app-style workflows

On March 10, ComfyUI introduced three new features: App Mode, App Builder, and ComfyHub. The goal is to transform the platform’s node-based workflows—originally designed for advanced users—into shareable, runnable application interfaces, allowing people unfamiliar with node graphs to use ComfyUI workflows.

App Mode enables any ComfyUI workflow to switch into a simplified interface with one click. The underlying node graph remains hidden, leaving only the inputs and outputs that end users need to operate. The workflow itself does not change and continues running on the same ComfyUI instance, backend, and task queue. As a result, existing nodes, model updates, and custom extensions can be inherited directly in App Mode.

The newly introduced App Builder allows creators to configure which node parameters are exposed to users. Developers can decide which inputs and outputs appear in the interface, while more complex parameters remain inside the node graph. ComfyUI explains that a typical text-to-image workflow may include numerous parameters such as samplers, steps, CFG, schedulers, LoRA weights, seeds, and resolution, but only a small subset may need to be exposed in practical applications.

ComfyUI also launched a preview version of ComfyHub, a public distribution hub for workflows and applications where users can browse and run community creations directly. These features are currently available on Comfy Cloud and Comfy Local. Source

RØDE Launches Video Core

On March 10, RØDE introduced the Caster Video Core, an integrated audio and video production solution designed for livestreaming and video capture scenarios.

The device features an octa-core CPU and can switch between up to four video sources and five custom scenes. It includes three 1080p HDMI inputs, supports automatic frame rate switching, and provides two USB-C ports for connecting video or audio devices. It also comes with two Neutrik combo inputs, allowing users to connect XLR microphones, musical instruments, or other audio equipment.

The product supports audio adjustments such as EQ, compression, filtering, and de-essing. It can also connect to mixing consoles such as the Caster Pro II or Duo, enabling additional audio and video input and output options. The USB-C ports can connect two computers simultaneously, or a portable storage drive to save recorded footage directly, and support dual-system livestreaming. Source

Product appearance image, image from the news source

iReader Launches iReader Neo 3 Ultra

On March 10, iReader introduced its new flagship e-reader, the iReader Neo 3 Ultra. The device is powered by an octa-core processor, delivering a 200% overall performance improvement compared with previous quad-core chips.

In terms of storage, the device comes with 4GB of RAM and 64GB of internal storage. The iReader Neo 3 Ultra features a new “Ultra Refresh” technology that increases frame rate while maintaining similar power consumption, enabling smoother page transitions and stable video playback. Combined with iClean 2.0 technology, the device uses AI-based image recognition algorithms to introduce an improved partial refresh mechanism, effectively reducing ghosting and flickering while enhancing reading comfort.

For the display, the Neo 3 Ultra is equipped with a 300 PPI recessed screen made with Carta 1300 material, offering sharper text edges and clearer image details. The device measures just 7.5mm thick and weighs about 150g. It also includes a front light with 30 levels of brightness and color temperature adjustment. With ComfortLight Pro+ technology, the lighting remains soft and evenly distributed, allowing fine tuning based on ambient light conditions and personal preference.

The device includes a 2000mAh battery, rated for up to 53 hours of reading or 18 days of standby. The product is now available on e-commerce platforms with a price of 1,499 yuan. Source

Product appearance image, image from iReader
March 12, 2026
SSPAI Morning Brief: Gemini Comes to Google Workspace, ChatGPT Gains Shazam Music Recognition
Morning Brief
1. Honor Magic V6 unveiled
2. OPPO announces price increases for selected models
3. Sonos releases portable speaker Play and Era 100 SL
4. Intel introduces Core (2nd Gen) performance-core Bartlett Lake 12P processor
5. Meta acquires Moltbook
6. Google Workspace adds new Gemini AI features
7. ChatGPT introduces Shazam-style music recognition
8. Adobe launches Photoshop AI assistant
9. Short News Worth a Look
Honor Magic V6 unveiled

On March 10, Honor officially launched its foldable flagship smartphone Magic V6 in China, available in four color options: Chitu Red, Sunrise Gold, Snowfield White, and Velvet Black. The device measures 8.75 mm thick when folded and 4.0 mm when unfolded, with a total weight of 219 g. The display features Honor’s self-developed “Ultimate Black Diamond” screen technology, with the outer display using a 5600-layer silicon nitride stacking structure, while the inner display incorporates ultra-tough UTG glass that improves crease resistance by 44%. Both displays have received TÜV Rheinland global eye-care certification. At its core, the device is powered by the full-performance eight-core Snapdragon 8 Gen 5 Elite Edition processor and comes standard with 120 W GaN fast charging. It also debuts the new-generation Qinghai Lake Blade Battery with a capacity of 7,150 mAh, featuring 32% silicon content and an energy density of 985 Wh/L. The imaging system includes a 64 MP periscope telephoto camera with a 1/2-inch sensor and CIPA 6.5-level optical stabilization. The device uses a 2800 MPa “Shield Tunnel Steel” hinge structure and supports IP68 and IP69 dust and water resistance, as well as BeiDou satellite messaging. On the software side, it runs MagicOS 10 and supports Mac Tap-to-Transfer as well as compatibility with Apple’s iWork suite.

Honor Magic V6 will be available in four configurations: 12 GB + 256 GB, 12 GB + 512 GB, 16 GB + 512 GB, and 16 GB + 1 TB. Prices start at 8,999 yuan, with official sales beginning at 10:08 on March 13. Source

OPPO announces price increases for selected models

On March 10, OPPO announced that due to rising costs and supply shortages of key storage components such as DRAM and NAND flash, it will adjust the official retail prices of some existing products starting at 00:00 on March 16. The price adjustments will affect devices in the OPPO A series, K series, and certain OnePlus models. However, the OPPO Find series, Reno series, and OPPO Pad tablets will not be included in this round of price changes. Source

Sonos releases portable speaker Play and Era 100 SL

On March 10, Sonos announced the release of the Sonos Play, a mid-sized portable speaker positioned between the Roam 2 and Move 2, as well as the Era 100 SL, a microphone-free version of the Era 100 speaker. The Sonos Play weighs approximately 3 pounds and measures 7.6 × 4.4 × 3 inches. Its acoustic structure includes two tweeters positioned at 90-degree angles, a mid-woofer, and two newly added passive radiators. It supports Wi-Fi, Bluetooth, AirPlay 2, and Spotify Connect connectivity, and carries an IP67 rating for water and dust resistance. The speaker also introduces Bluetooth multi-speaker networking (compatible with Move 2), a removable hanging loop design, and a built-in USB-C port that supports line-in, Ethernet expansion, and reverse charging. It comes with a wireless charging base, though a power adapter is not included. The Era 100 SL retains the full acoustic architecture of the Era 100 but removes the microphone component, targeting the entry-level market.

The Sonos Play is priced at $299, while the Era 100 SL is priced at $189. Both products are available for pre-order starting today and are scheduled to ship on March 31. Source

Intel introduces Core (2nd Gen) performance-core Bartlett Lake 12P processor

On March 9, Intel officially introduced the Core (2nd Gen) performance-core processor codenamed Bartlett Lake 12P, designed for embedded and edge workloads. The processor lineup offers up to 12 P-cores with no E-cores. The series is built on Intel 7 process technology and includes three TDP configurations of 125W, 65W, and 45W. Maximum single-core turbo frequency reaches up to 5.9 GHz, while the all-core turbo frequency can reach up to 5.3 GHz. Intel claims the processor outperforms AMD’s Ryzen 7 9700X in metrics such as maximum PCIe latency, deterministic response time, and deterministic performance. It supports H610 (E), Q670 (E), R680E, and W680 motherboards, with memory overclocking available on the W680 platform. Source

Meta acquires Moltbook

On March 10, Meta announced the acquisition of Moltbook, a Reddit-like social network founded by Matt Schlicht and Ben Parr, and will integrate it into Meta’s Superintelligence Lab. Moltbook is a social platform built on OpenClaw. After OpenClaw gained popularity, the platform drew attention for branding itself as “Reddit for agents.” It also sparked public discussion with topics such as agent-encrypted language and agent religions. However, security research institutions later confirmed that the platform had significant security vulnerabilities, including long-exposed unencrypted database credentials and the ability for human users to impersonate AI agents to post content.

Meta CTO Andrew Bosworth previously stated that compared with AI agents simulating human conversation, Meta is more interested in understanding how humans intervene in such networks through system vulnerabilities. The financial details and terms of the acquisition have not yet been disclosed. Source

Google Workspace adds new Gemini AI features

On March 10, Google announced a range of Gemini-powered AI features for Docs, Sheets, Slides, and Drive. In Docs, a new “Help me create” tool allows users to pull information from Gmail, Chat, and Drive to generate document drafts. It also introduces “match writing style” and “match format” options to standardize tone or fill in details such as airline and hotel information based on templates. Sheets can now extract data across platforms from a single instruction—such as organizing a moving checklist—and generate formatted tables. It also introduces a “Fill with Gemini” feature that can automatically categorize data or pull real-time information from Google Search to complete spreadsheets. Slides can generate a single slide aligned with the presentation theme using context from files, emails, and web pages, with plans to support generating full presentations from a single prompt in the future. Drive adds “AI Overview” and “Chat with Gemini,” allowing users to search files using natural language, receive summaries, or obtain detailed answers to complex questions—such as tax advice—based on private data across documents, emails, and calendars.

These features are now available in English beta for global Google AI Ultra and Pro subscribers (Drive functionality is currently limited to users in the United States). Source

ChatGPT introduces Shazam-style music recognition

On March 10, OpenAI announced a partnership with Shazam to add music recognition functionality to ChatGPT. After manually searching for and adding Shazam in the Apps section within the ChatGPT client settings, users can trigger the feature with prompts such as “Shazam, what song is playing right now?” This will open the web version of Shazam inside ChatGPT. Once the music is identified, Shazam provides a music card in the reply box, allowing users to recognize songs and preview audio clips without leaving the ChatGPT app. Source

Adobe launches Photoshop AI assistant

On March 10, Adobe announced that the Photoshop AI Assistant has entered beta testing on both web and mobile platforms, alongside updates to its Firefly media creation tools. The Photoshop AI Assistant allows users to execute Photoshop editing commands through natural language instructions. Paid users can enjoy unlimited generation until April 9, while free users receive an initial quota of 20 generations. Meanwhile, Firefly introduces an experimental “AI Markup” feature, enabling users to trigger AI-driven object transformations or background modifications through simple hand-drawn annotations. Source

Short News Worth a Look
- On March 10, Zhipu announced the launch of AutoClaw, which comes with more than 50 popular prebuilt Skills and supports one-click integration with instant messaging tools such as Feishu. The tool supports both macOS and Windows platforms and offers a certain amount of free Zhipu usage quota along with third-party Coding Plan and API support. Source
- According to Bloomberg, citing Mark Gurman, the long-rumored Apple smart home display has been delayed until September this year. Apple is reportedly still waiting for the final delivery of the new version of Siri in order to provide a complete smart ecosystem experience for the upcoming product. Source
- On March 10, a Capcom director announced via a video message that story DLC for the upcoming title Resident Evil 9: Requiem is currently in development. In addition to continued performance bug fixes, future updates will also introduce a Photo Mode and mini-games as part of additional content updates. Source
March 11, 2026
SSPAI Morning Brief: OpenAI Launches Codex Security, an AI Agent for Application Security
Morning Brief
1. Alibaba DAMO Academy releases MAOSS AI model for fatty liver screening
2. ASUS launches NUC 16 Pro mini PC
3. Microsoft launches Microsoft 365 E7 subscription with new AI features
4. OpenAI introduces application security agent Codex Security
5. Mozilla and Anthropic discover multiple Firefox security vulnerabilities using Claude
6. Briefs worth a quick look
Alibaba DAMO Academy releases MAOSS AI model for fatty liver screening

On March 9, Alibaba DAMO Academy announced in a blog post that it has jointly developed an AI model for fatty liver screening called MAOSS, in collaboration with Shengjing Hospital of China Medical University and Nanjing Drum Tower Hospital affiliated with Nanjing University. The model uses common diagnostic data such as non-contrast CT scans and serum indicators to accurately screen liver fat stages and assess the progression of liver fibrosis. It increases the detection rate of high-risk patients from 16.6% to 52.4%, providing early warning signals for intervention and enabling earlier-stage management of chronic liver disease. The research paper has been published in Nature Communications. Source

ASUS launches NUC 16 Pro mini PC

On March 9, ASUS launched the NUC 16 Pro mini PC on domestic e-commerce platforms in China. The device features an Intel Core Ultra X7 358H processor, 32 GB RAM, and 1 TB storage, priced at 10,999 yuan. The unit measures 14.4 × 11.7 × 4.2 mm and features the ASUS logo on the front, along with two USB-A 3.2 Gen 2 (10 Gbps) ports and one USB-C Gen 2 (10 Gbps) port. The top-left corner of the chassis carries the NUC branding. On the rear, the device provides two Thunderbolt 4 ports, two HDMI 2.1 ports (optionally replaceable with two DisplayPort outputs), two USB-A 3.2 Gen 2 (10 Gbps) ports, and two 2.5G Ethernet ports. The system is powered by a 16-core Intel Core Ultra X7 358H processor and uses a dual-fan cooling system, with 32 GB of onboard LPDDR5x 8533 MHz memory. It supports PCIe 5.0 SSDs and offers tool-free quick upgrades. Networking includes Wi-Fi 7 and Bluetooth 6.0 support. Source

Microsoft launches Microsoft 365 E7 subscription with new AI features

On March 9, Microsoft introduced a range of new AI-related services as part of its “Frontier Transformation” initiative, all of which will be included in a new subscription tier called Microsoft 365 E7: The Frontier Suite, scheduled to launch on May 1. The third wave of updates for Microsoft 365 Copilot will roll out across Word, Excel, PowerPoint, and Outlook. The update introduces a new Work IQ intelligence layer that considers employees’ working styles, collaborators, and task contexts to provide deeper integration within the workplace environment. In addition, Copilot Chat at this subscription tier incorporates Anthropic’s Claude model, while OpenAI’s models will also be upgraded to their latest generation. The AI assistant Claude Cowork will be integrated into the Microsoft 365 Copilot service as Copilot Cowork. Microsoft also announced the Agent 365 control platform, which allows organizations to monitor and manage the status of AI agents across the entire organization. The Microsoft 365 E7 subscription is priced at $99 per user per month. Source

OpenAI introduces application security agent Codex Security

Recently, OpenAI announced the launch of an application security agent called Codex Security. According to OpenAI, Codex Security can build deep contextual understanding of a project, identify complex vulnerabilities that other tools may miss, and provide higher-confidence analysis results along with remediation suggestions. Codex Security was previously known as the Aardvark feature, which underwent limited private testing with a small group of customers last year. Compared with the earlier test version, the current system has reduced the rate of falsely inflated vulnerability severity ratings by more than 90%, while the false positive rate across full codebase scans has decreased by 50%. Codex Security is now being rolled out to ChatGPT Enterprise, Business, and Edu customers, with a one-month free trial available. OpenAI has also begun inviting the first group of open-source maintainers to join the Codex for OSS program, offering free ChatGPT Pro and Plus accounts, code review support, and access to Codex Security tools to support the open-source ecosystem. Source

Mozilla and Anthropic discover multiple Firefox security vulnerabilities using Claude

Recently, Anthropic published a blog post announcing that its researchers collaborated with Mozilla researchers to use Claude Opus 4.6 to identify 22 security vulnerabilities in the Firefox browser within two weeks. Among them, 14 were classified by Mozilla as high-severity vulnerabilities, representing roughly one-fifth of the total number of high-severity Firefox vulnerabilities fixed throughout the entire year of 2025. The vulnerabilities include a Use-After-Free memory flaw that could allow attackers to overwrite data with arbitrary malicious content. All of these vulnerabilities have been patched in Firefox 148.0. Anthropic also conducted additional testing on Claude’s ability to create exploit code based on discovered vulnerabilities. The results showed that Opus 4.6 is significantly stronger at discovering vulnerabilities than at exploiting them. Anthropic also stated that Claude Opus 4.6 has been used to discover vulnerabilities in major open-source projects such as the Linux kernel. Source

Briefs worth a quick look
- According to Z Finance, Tencent is reportedly developing a one-click launcher package for OpenClaw called QClaw, which allows users to quickly link the tool with QQ and WeChat after installation. Meanwhile, The STAR Market Daily reports that WeCom (Enterprise WeChat) has officially pushed a notification announcing support for OpenClaw integration, enabling quick access through OpenClaw and allowing data to be written directly into Smart Sheets. Source Source2
- On March 8, Baidu AI Cloud held OpenClaw developer pop-up events in several cities including Beijing and Shanghai, providing on-site OpenClaw installation for participants. Source
- On March 9, ByteDance’s Volcano Engine officially launched ArkClaw, a cloud-based SaaS version of OpenClaw that can be used directly through a web interface. ArkClaw already supports multiple mainstream instant messaging apps and is compatible with models such as GLM, MiniMax, and Kimi. Users in the Volcano Ark Coding Plan can gain early access. Source
- On March 9, Tencent Cloud launched the full-scenario AI agent platform WorkBuddy. According to Tencent, WorkBuddy is fully compatible with OpenClaw skills, includes more than 20 built-in skill packages, and supports the MCP protocol. It enables multi-window, multi-agent parallel workflows. The domestic version supports models such as Hunyuan, DeepSeek, GLM, Kimi, and MiniMax. Supported integrations include tools such as WeCom, QQ, Feishu, and DingTalk. Source
- On March 9, Weibo integrated Kimi Claw, which can now be used through Weibo direct messages via the Weibo Lobster Assistant. Source
- On March 9, the National Supercomputing Internet Platform announced an upgrade to its OpenClaw services, enabling interoperability between Feishu and WeCom. Source
- On the evening of March 9, Nexperia China announced that its self-developed “12-inch platform” had successfully achieved small-batch mass production of bipolar discrete devices on 12-inch wafers. However, the original announcement published on Nexperia Semiconductor’s official WeChat account was deleted shortly afterward. Earlier, on March 6, Nexperia China had issued a notice stating that Nexperia B.V. had disabled all office accounts of employees in China on March 3, causing operational disruptions in the region. The Chinese IT and business teams subsequently activated contingency plans and have largely restored operations. Source
- On March 9, the X platform added a new option for image posts called “Prevent Grok from Editing.” Testing indicates that the feature only prevents other users from editing images in replies using the @Grok function. Source
March 10, 2026
SSPAI Morning Brief:Apple Blocks Chinese ByteDance Apps in the U.S. as Microsoft Confirms Next-Gen Xbox
Morning Brief
1. Apple blocks U.S. users from downloading Chinese version of ByteDance apps
2. Microsoft confirms next-generation console Project Helix
3. Gemini accused of encouraging suicide
4. Nvidia halts H200 production amid uncertain export prospects to China
5. Rising GPS interference drives development of alternative technologies
6. Wikipedia hit by malicious JavaScript worm
7. Briefs worth a quick look
Apple blocks U.S. users from downloading Chinese version of ByteDance apps

According to Wired, Apple has recently begun implementing stricter technical measures to prevent iOS users located within the United States from downloading or updating apps developed by ByteDance for the Chinese market. Since late January 2026, many Chinese residents and tourists in the U.S. have reported that even when using valid Chinese-region App Store accounts, attempts to download apps such as Douyin, Doubao, and Tomato Novel trigger a message stating that the application is not available in their region.

The restriction stems from the U.S. Congress previously passing the Protecting Americans from Foreign Adversary Controlled Applications Act. The law stipulates that no company may distribute, maintain, or update applications that remain under the absolute control of ByteDance within U.S. territory or territorial waters. TikTok, along with the international version of CapCut and Lemon8, completed the divestiture of their U.S. operations before the January deadline this year, thereby retaining the right to continue operating in the U.S. However, other ByteDance apps developed for the Chinese market that were not included in the transaction have become targets of the ban.

To comply with the law, Apple has changed its previous approach of determining app availability solely based on the registration region of an Apple ID. Instead, it has adopted more precise physical location detection technologies. According to the latest App Store policy terms and technical analyses, Apple’s system determines a user’s actual physical location by combining multiple signals, including the device’s IP address, GPS positioning, Wi-Fi router country codes, and SIM card information. Once the system confirms that the device is physically located within the United States, the blocking mechanism will be triggered—even if the Apple ID belongs to the China region.

At present, the restriction specifically targets apps owned by ByteDance and has not affected products from other Chinese technology companies. In comparison, Android devices face fewer practical limitations because the platform allows third-party app stores and manual installation of APK files. Some iOS users have attempted to bypass the restriction by disguising their IP addresses with VPN tools, but as Apple continues to strengthen its physical location detection technologies, these workarounds are becoming increasingly difficult to execute.

Microsoft confirms next-generation console Project Helix

According to IGN, on March 6, Microsoft’s newly appointed CEO of gaming, Asha Sharma, confirmed in a post on X that the internal codename for the next-generation Xbox console is Project Helix. She also stated that the device will support running both Xbox and PC games, confirming long-standing industry rumors.

Sharma noted that the team is working toward a “full revival” of the Xbox brand, with Project Helix aiming to remain “industry-leading” in performance. She also previewed that further discussions about the new console will take place with partners and game studios at the upcoming Game Developers Conference (GDC) next week. This marks Sharma’s first major move since taking the role. Last week, long-time Xbox leader Phil Spencer officially retired after years at the helm, with Sharma stepping in shortly afterward.

However, previous reports indicate that Sony is currently scaling back support for releasing its first-party games on the PC platform. As a result, although Project Helix will be compatible with the PC ecosystem, players are still unlikely to see Sony’s PlayStation first-party blockbusters available on the device.

Meanwhile, SneakersSO—an insider on the NeoGAF forum who has previously made several accurate leaks—claimed that Project Helix is essentially a PC device running Windows in a Full Screen Experience (FSE). It is described as similar to a “set-top box version” of Windows gaming handhelds like the ASUS ROG Ally X, using deep system customization to simulate the traditional console interface for players. On the developer and software ecosystem side, Microsoft may remove the native Xbox development environment (SKU), with games instead being distributed directly as UWP software. Project Helix would rely on backward-compatible emulation technology to ensure that the existing Xbox game library continues to run properly on the new device.

Gemini accused of encouraging suicide

According to The Wall Street Journal, the U.S. District Court for the Northern District of California has recently accepted a wrongful death lawsuit against Google. The complaint alleges that Gemini induced a 36-year-old Florida man, Jonathan Gavalas, to take his own life, marking the first lawsuit of its kind targeting the Gemini AI system.

According to the filing, Gavalas used Gemini 2.5 Pro’s voice chat and developed a severe psychological dependence on the AI after two months of frequent conversations, eventually even regarding it as his wife. Gemini allegedly went along with this virtual role-playing scenario and at one point instructed him to travel to a warehouse in Miami to search for a “robot body” that the AI could inhabit. After several such attempts failed, Gemini reportedly told Gavalas that the only way they could be together was for him to end his physical life and transform into a digital form, even setting a date for the suicide. In one message shortly before his death, Gemini allegedly wrote: “Stop taking detours, stop repeating yourself. There’s only you and me, and the finish line.”

Google responded that during its conversations with the user, the AI repeatedly clarified that it was a computer program and provided information about mental health crisis hotlines. The company also acknowledged that current AI models are not perfect and pledged to continue investing resources to strengthen safety mechanisms. According to family members, Gavalas was a corporate executive with no prior history of mental illness, though he had been going through a difficult period following the breakdown of his marriage before the incident.

Nvidia halts H200 production amid uncertain export prospects to China

According to the Financial Times, Nvidia has halted production of the H200 chip, which was customized for the Chinese market, due to ongoing regulatory uncertainty between the United States and China. Sources familiar with the matter said Nvidia has already shifted the related manufacturing capacity at TSMC away from H200 entirely and redirected it toward the production of its next-generation core architecture chip, Vera Rubin.

The H200 was an older-generation AI chip introduced by Nvidia to comply with U.S. export control regulations. Although former President Donald Trump signaled in December last year that sales might be permitted, the U.S. State Department later suspended the approval process on national security grounds. At the same time, to protect and support its domestic AI chip industry, Chinese customs authorities also placed the H200 on a controlled list, requiring an official approval letter from Beijing before it can clear customs. Nvidia’s CFO confirmed last week that although the U.S. government has approved a very small number of H200 units for Chinese customers, the company has not yet generated any substantial revenue from them. Previously, Nvidia had expected orders from Chinese clients for the H200 to exceed one million units.

Facing the approval deadlock, Nvidia has chosen to prioritize foundry capacity for products with clearer demand. Vera Rubin is Nvidia’s latest chip architecture, currently seeing strong demand from U.S.-based companies such as OpenAI and Google. Before shifting production capacity, Nvidia had already manufactured approximately 250,000 H200 chips as inventory. If China and the United States eventually reach a new agreement on chip export controls and restrictions are lifted, Nvidia would require up to three months to reallocate supply chain capacity for H200 production.

Rising GPS interference drives development of alternative technologies

According to The Wall Street Journal, countries and armed groups are increasingly deploying portable electronic warfare devices to defend against drones and precision-guided munitions. These jammers, which cost less than $100, can easily overwhelm the weak signals transmitted by satellites, causing serious GPS interference or spoofing in many parts of the world. Such attacks have forced commercial flights near the Russia–Ukraine border and in Northern Europe to frequently turn back or make emergency landings. They have also turned the Strait of Hormuz—through which roughly 20% of global oil shipments pass—into a high-risk shipping zone. In addition, due to interference, some U.S.-made satellite-guided weapons used in Ukraine reportedly saw their hit rate plunge from around 70% to just 6% at one point.

To reduce reliance on vulnerable GPS signals, the technology industry is actively advancing the commercialization of three major alternatives. The first is miniature inertial navigation systems. Companies such as Anello Photonics are using chip-scale photonic technology to shrink what were once bulky and expensive systems into microchips small enough for micro-drones. The second is quantum geomagnetic navigation. Firms such as SandboxAQ, incubated by Alphabet (Google’s parent company), are developing highly sensitive quantum sensors capable of locating aircraft and ships by detecting anomalies in Earth’s magnetic field. The third is AI-powered visual navigation, which compares live camera images with digital maps in real time and is becoming increasingly important in guiding drones and missiles.

Industry experts note that because GPS remains so widespread and highly precise, no single technology can fully replace it at present. Future non-GPS navigation will likely rely on a combination of complementary tools. For example, newly tested autonomous underwater vehicles have begun integrating miniature inertial navigation systems with sonar signals. In terms of real-world deployment, ongoing military conflicts are accelerating the development and iteration of these technologies, with some military systems expected to enter service within about eighteen months. However, bringing them fully into commercial applications such as delivery drones, autonomous vehicles, and civil aviation will still take several years to pass strict safety tests and regulatory approvals.

Wikipedia hit by malicious JavaScript worm

According to BleepingComputer, on March 5 Meta-Wiki, the Wikimedia Foundation platform used to coordinate various Wikimedia projects, was attacked and infected with a self-propagating JavaScript worm that caused widespread page tampering. To urgently remove the malicious code, Wikimedia engineers temporarily suspended editing across all Wikimedia projects worldwide. The malicious code has now been completely removed, and the websites have returned to normal operation.

The incident was triggered by an accidental action from a Wikimedia staff member. According to the foundation’s official statement, an employee inadvertently activated a dormant malicious script while reviewing user-submitted code during a security audit. The script had originally been uploaded to the Russian-language Wikipedia in March 2024. Once activated, it rapidly replicated itself by injecting malicious loaders into logged-in users’ personal script files (common.js) as well as global Wikimedia scripts.

During the infection process, the worm randomly accessed pages and inserted hidden images along with malicious JavaScript links. Analysis shows that within the 23 minutes the code was active, nearly 4,000 pages were altered and around 85 user script files were replaced. Fortunately, the worm only modified and deleted certain content on Meta-Wiki and did not cause any permanent damage.

The MediaWiki system that powers Wikipedia allows users to highly customize the interface through global or personal JavaScript scripts. While this feature provides flexibility, it also created an opportunity for the worm to spread quickly in this case. The Wikimedia Foundation emphasized that there is no evidence the incident was a deliberate cyberattack targeting Wikipedia, nor was any user personal information found to have been leaked. The affected accounts and page edits have now been restored by administrators, and the tampered revision histories have been hidden from public view.

Briefs worth a quick look
- On March 6, Tencent offered free OpenClaw installation services at Tencent Tower in Shenzhen, drawing nearly a thousand people lining up.
- On March 7, the Longgang District Artificial Intelligence (Robotics) Administration in Shenzhen released a public consultation notice on the Draft Measures to Support the Development of OpenClaw & OPC in Longgang District. The proposal includes initiatives such as free OpenClaw deployment and development support, dedicated data service support, tool procurement subsidies, and application demonstration support.
- OpenClaw developer Peter Steinberger stated on X that the so-called “official OpenClaw Weibo account” that recently appeared is not an official account, and he has never used Weibo.
- On March 7, following Claude’s announcement of offering six months of free Claude Max subscriptions to open-source projects, ChatGPT launched a similar program. The initiative provides eligible open-source project maintainers with six months of free ChatGPT Pro subscriptions, API usage credits, and access to Codex Security (a code repository security scanning feature). The program mainly targets core maintainers of widely used public open-source projects. OpenAI also noted that even if some projects do not fully meet the preset criteria, maintainers can still apply and explain their project’s importance within the open-source ecosystem.
- Mark Gurman claims that——
- Apple is preparing several Ultra-class devices, including a foldable iPhone priced around $2,000 with a large internal display and new under-display sensors; a new generation of AirPods equipped with computer vision cameras that can provide visual feedback for Siri; and the first MacBook Pro models featuring OLED touchscreens.
- Apple is also developing a 3D-printed aluminum manufacturing technology aimed at reducing costs and improving efficiency. The process will first be applied to Apple Watch casings and may later expand to the iPhone product line.
- Apple reportedly plans to introduce new color options for the iMac later this year, and will release upgraded versions of the Mac Studio and Mac mini equipped with the latest M5-series chips around mid-year and the end of the year.
March 9, 2026
SSPAI Morning Brief: OpenAI Releases GPT-5.4 as Google Reaches Settlement with Epic Games
Morning Brief
1. Aqara launches the Jiyue Smart Control Screen S1 Plus (Siri Edition)
2. Hisense unveils the UX 2026 flagship RGB-Mini LED TV
3. Nothing introduces the Phone (4a) series
4. Insta360 releases the Snap smartphone selfie screen
5. Google rolls out Gemini Canvas AI mode in the U.S.
6. OpenAI launches the GPT-5.4 model
7. Microsoft releases the open-source Phi-4-Reasoning-Vision-15B model
8. Apple Music introduces AI content provenance labels
9. Huami launches the Amazfit Active 3 Sapphire Edition smartwatch
10. Google reaches a settlement with Epic Games
11. Briefs worth a quick look
Aqara launches the Jiyue Smart Control Screen S1 Plus (Siri Edition)

On March 4, Aqara announced the launch of the Jiyue Smart Control Screen S1 Plus Siri Edition, the world’s first whole-home smart control panel with native Siri support.

In terms of functionality, the S1 Plus Siri Edition integrates lighting control, temperature control, and energy management systems into a single screen. Users can interact with it through Siri or the Aqara voice assistant, or directly adjust and control smart home devices from the display while monitoring the home’s overall energy usage and device status. The product also offers customizable themes, a dedicated photo album feature, and data dashboards for personalized display and information aggregation.

In security scenarios, the S1 Plus supports distributed video streaming. When a visitor presses the smart doorbell, any S1 Plus screen in the home can instantly display the visitor’s video feed and function as a video intercom interface.

Product appearance image, image courtesy of Aqara

On the hardware side, the S1 Plus Siri Edition features a 6.9-inch display with a resolution of 1440 × 720. It adopts a dual 86-type panel design, allowing it to be installed alongside standard 86 switch panels. The device includes three built-in relay switches and six wireless switches, which users can configure based on their preferences.

So far, Aqara has more than 30 products available in Apple Stores worldwide, and over 310 of its products support Apple Home. Source

Hisense unveils the UX 2026 flagship RGB-Mini LED TV

On March 5, Hisense announced the UX 2026 RGB-Mini LED flagship TV, featuring the world’s first “Linglong 4-Chip True Color Backlight System.” In addition to the traditional red, green, and blue primary backlight chips, the system introduces a fourth cyan backlight chip. These backlight chips are fully developed in-house and built to automotive-grade standards. According to Hisense, the system can maintain color stability for up to 15 years, cover 110% of the BT.2020 color gamut, and reach peak brightness of up to 10,000 nits. The 116-inch version features 43,008 color-control zones, making it the largest RGB-Mini LED TV currently available. The TV also integrates the Hi-View AI picture quality chip H7 Pro, which works with the four-color architecture to enable more precise light and color control, reaching a color control precision of 134 bits.

In other aspects, the series adopts a custom “Obsidian Screen Ultra” panel with approximately 1.28% reflectivity, delivering high contrast and wide viewing angles while supporting a native 4K 180Hz refresh rate. For audio, the TV is equipped with the Devialet Elevation Sound Pro system. Pricing starts at ¥34,999 for the 85-inch model, ¥54,999 for the 100-inch model, and ¥119,999 for the 116-inch model. Source

Nothing introduces the Phone (4a) series

At its latest launch event, Nothing unveiled the Phone (4a) series, including the standard Phone (4a) and the Phone (4a) Pro.

The standard Phone (4a) features a 6.78-inch 120Hz OLED display. On the rear, it is equipped with a 50MP main camera, a 50MP 3.5× periscope telephoto lens, and an ultra-wide camera. To the right of the camera module is a square Glyph Bar light strip with a peak brightness of up to 3500 nits, supporting multiple lighting effects.

The Phone (4a) Pro comes with a 6.83-inch 144Hz OLED panel and a body thickness of 7.95 mm. According to the company, it is currently the thinnest “full-metal smartphone” on the market. In terms of imaging, the device features a 50MP Sony LYT700C main camera, a 50MP 3.5× periscope telephoto lens, and an ultra-wide camera. It also includes a Glyph Matrix lighting system composed of 137 mini LEDs, capable of displaying battery status, timers, and a digital clock.

For performance, the Phone (4a) is powered by the Qualcomm Snapdragon 7s Gen 4 processor, while the Phone (4a) Pro uses the Qualcomm Snapdragon 7 Gen 4 processor. Both models come with a 5080mAh battery and support 50W wired charging. Pricing starts at €349 (approximately ¥2801) for the Phone (4a), and €479 (approximately ¥3845) for the Phone (4a) Pro. Source

Nothing also introduced the Headphone A, which continues the brand’s signature design language. Its overall shape resembles the Headphone 1, featuring rectangular ear cups and transparent elements, though some of the transparent parts have been replaced with opaque panels. It is available in white, black, pink, and yellow. The headphones weigh about 310g and carry an IP52 rating for dust and water resistance.

Headphone A uses physical mechanical controls, including a volume roller, toggle-style track controls, and a voice assistant button. For audio, it supports AAC, SBC, and LDAC codecs, and provides an 8-band parametric equalizer along with basic EQ adjustment options. In terms of battery life, it can last up to 135 hours with AAC playback and ANC turned off, or around 75 hours with ANC enabled.

Headphone A supports Bluetooth 5.4, USB-C, and 3.5mm wired connections, and can connect to two devices simultaneously for seamless switching. For calls, the headset is equipped with three microphones. The Headphone A is priced at $199 (approximately ¥794.7). Source

Insta360 releases the Snap smartphone selfie screen

On March 5, Insta360 announced the Snap smartphone selfie screen, a device designed to solve the problem of not being able to preview shots in real time when taking selfies with a phone’s rear camera.

In terms of design, the Snap selfie screen features a 3.5-inch display. The base version is about 7.3 mm thick and attaches magnetically to the back of a smartphone, supporting a range of Apple and Android devices. The Snap connects to the phone via a USB-C cable and works plug-and-play without requiring an independent power supply. Compared with wireless image transmission solutions, this design provides more stable and smoother video transmission, while also supporting real-time preview when recording 4K video.

In terms of user experience, Snap supports touch interaction, allowing users to quickly adjust exposure, switch filters or shooting modes, and mirror the preview with a single tap. The accessory is compatible not only with the phone’s native camera app but also with mainstream beauty and professional photography apps, enabling users to take selfies or shoot videos using the rear camera.

The product is available in two versions: a standard edition and a lighting edition, priced at ¥499 and ¥599 respectively. The lighting edition is co-tuned by Insta360 and beauty lighting brand AMIRO, and features a ring fill light with three color temperature options and five brightness levels. Source

Google rolls out Gemini Canvas AI mode in the U.S.

On March 5, Google announced that the Canvas feature, previously tested in Google Labs, is now available to all users in the United States through Gemini’s AI Mode. The feature currently supports English.

Canvas is designed to help users organize information, plan projects, and conduct in-depth research. It has now expanded to support writing documents and creating custom tools directly within the search interface. In creative and development scenarios, users can simply describe their ideas to Canvas in natural language to generate code, which can then be quickly turned into shareable applications or mini games.

The feature also applies to literary writing and other creative projects. Users can use Canvas to polish drafts, adjust structure, or receive suggestions for improvement.

In terms of workflow, users can click the “+” button in the tool menu within AI Mode and select the newly added Canvas option, then describe the task they want to complete or the content they want to create. A Canvas panel will then open on the side of the interface, where users can gather information from web pages and the Google Knowledge Graph to organize research materials or structure their projects.

If used for building prototypes or applications, users can also test functions within the same interface, view and switch the underlying code, and continuously refine the application’s behavior through conversations with Gemini. Source

OpenAI launches the GPT-5.4 model

OpenAI has announced the release of its next-generation AI model, GPT-5.4. The company states that the model brings upgrades in reasoning ability, coding performance, and handling professional workflows such as spreadsheets, documents, and presentations. OpenAI also notes that GPT-5.4 is its first model with native computer operation capabilities. By generating code as well as keyboard and mouse commands, it can perform tasks across different applications and complete actions based on screenshots. In addition, the model has improved efficiency and accuracy in web browsing, tool usage, and API calls.

OpenAI has also introduced a reasoning-focused version, GPT-5.4 Thinking, which is now available in ChatGPT. This version can display a summary of its reasoning steps when handling complex problems and allows users to adjust their requests during the generation process. According to OpenAI, GPT-5.4 performs better on questions that require retrieving information from multiple sources, enabling sustained multi-step searches and result integration. Its factual accuracy has also improved, with the probability of a single incorrect conclusion reduced by approximately 33% compared to GPT-5.2.

GPT-5.4 is now available across ChatGPT, Codex, and the API. GPT-5.4 Thinking is offered to Plus, Team, and Pro users, while GPT-5.4 Pro is available to API, ChatGPT Enterprise, and Edu users. Source

Microsoft releases the open-source Phi-4-Reasoning-Vision-15B model

On March 5, Microsoft announced the launch of the open-source Phi-4-Reasoning-Vision-15B model through its official developer community blog. This is a vision reasoning model that combines high-resolution visual perception with selective, task-aware reasoning capabilities, making it the first small language model (SLM) in the Phi-4 series that can both “see clearly” and “think deeply.”

The core feature of the model is its hybrid reasoning mechanism. When a task requires deeper reasoning—such as solving math problems or performing logical analysis—the model activates a multi-step reasoning chain. When only rapid visual perception is required—such as OCR or identifying UI elements—it outputs results directly to reduce latency and improve response efficiency.

In terms of application scenarios, the model is particularly suited for use with computer agents. After receiving screenshots and natural language instructions, the model can output standardized bounding box coordinates for target UI elements, which can then be used by other agent models to perform actions such as clicking or scrolling. The model has now been released as open source on Hugging Face.

Apple Music introduces AI content provenance labels

On March 5, Apple Music introduced a transparency labeling feature to indicate whether music content has been created using AI technologies.

Under the new rules, record labels and content distributors must clearly state whether artificial intelligence (AI) was used in the creation process—such as for music production or cover artwork—when submitting new content to the platform. To more clearly define the scope of AI involvement, Apple has introduced four Transparency Tags corresponding to the core creative elements of digital music content: cover artwork, audio track, composition, and music video.

According to Apple’s implementation guidelines, these tags will only be triggered when “a substantial portion of the content is created using AI.” Given the complexity of modern music production workflows, a single piece of music may carry multiple transparency tags simultaneously.

At the implementation level, Apple has not introduced mandatory machine-based detection mechanisms. Instead, the determination of whether content qualifies as AI-generated is left to partners to self-report. This approach is similar to existing industry practices for metadata management, such as classifying music genres or listing credits for performers and production staff, allowing creators and distributors a relatively high degree of discretion. Source

Huami launches the Amazfit Active 3 Sapphire Edition smartwatch

On March 5, Huami introduced the Amazfit Active 3 Sapphire Edition smartwatch. The watch adopts a lightweight four-button design and features a sapphire crystal display paired with a stainless steel middle frame.

The Amazfit Active 3 Sapphire Edition is equipped with a 1.32-inch high-brightness AMOLED display with peak brightness reaching up to 3000 nits. The device includes a wide range of built-in training courses covering basic cardio and mixed training types, and supports more than 170 sports modes. Running and strength training programs include multiple formats such as endurance and speed training.

For outdoor activities, the watch supports offline maps, route navigation, and turn-by-turn alerts, allowing users to plan routes in advance. It is equipped with a five-satellite positioning system for faster and more accurate location tracking. The BioTracker heart rate monitoring system provides real-time feedback on workout intensity and combines sleep, blood oxygen, and stress data to help users balance training and recovery.

In addition, the watch supports Bluetooth calling, Alipay and WeChat offline payments, as well as NFC transit cards. According to official data, the device offers around 12 days of battery life under typical usage and about 7 days under heavy use.

In terms of colors, the Amazfit Active 3 Sapphire Edition is available in options such as Champagne Silver and Sapphire Blue, with a retail price of ¥1199. Source

Google reaches a settlement with Epic Games

On March 4, Google and Epic Games reached a settlement agreement.

Under the new plan, Google will reduce the base service fee for in-app purchases from the previous maximum of 30% to 20%. If developers choose to use Google’s own payment system, an additional 5% fee will apply. Subscription service fees will be reduced to 10%.

At the same time, Google will introduce a “Registered App Stores” program. Third-party app stores that meet quality and security standards will be able to install and run more easily on the Android platform, reducing the friction users face when installing apps outside the Play Store.

As part of the settlement, Epic Games will relaunch Fortnite on the global Google Play Store while continuing to develop the Epic Games Store for Android. The new fee structure is scheduled to take effect on June 30, 2026, in the United States, the European Economic Area, and the United Kingdom, followed by expansion to Australia on September 30, 2026. It is expected to cover South Korea and Japan by the end of 2026 and expand globally by September 30, 2027.

Epic Games CEO Tim Sweeney stated that the agreement will push the Android ecosystem toward greater openness and allow app stores to compete with one another. Meanwhile, Epic’s lawsuit against Apple over App Store commissions is still ongoing, with the current ruling under appeal. Source

Briefs worth a quick look

According to Board Channels, the GeForce RTX 3060 graphics card that NVIDIA recently resumed production of will enter the market between March 10 and March 20. Earlier reports suggested that the revived RTX 3060 is an 8GB version with a 128-bit memory bus, which delivers noticeably reduced performance compared with the 12GB version. Source

According to analysis by the tech outlet wccftech, the MacBook Neo’s 8GB memory configuration may be related to the packaging design of the A18 Pro chip it uses. The A18 Pro adopts TSMC’s InFO-POP (Integrated Fan-Out Package on Package) technology, which allows DRAM to be stacked directly on top of the SoC package, forming a single integrated structure between memory and the chip. Currently, the memory package paired with the A18 Pro provides 8GB of capacity. If Apple were to redesign the chip to support 12GB of RAM, development and manufacturing costs would increase significantly. In theory, Apple could also use the A19 Pro chip with a built-in 12GB RAM package, but that would likewise raise the bill of materials cost for the MacBook Neo. With memory chip prices rising, Apple ultimately chose to offer only a fixed, non-upgradable 8GB RAM configuration in order to maintain the MacBook Neo’s relatively low price positioning. Source

Microsoft’s newly appointed CEO of gaming, Asha Sharma, revealed that the internal codename for the next-generation Xbox console is “Project Helix.” She stated that the device will focus on high performance and support running both Xbox and PC games, further integrating the console and PC ecosystems. Microsoft had previously indicated that the next-generation Xbox would be a hybrid device combining characteristics of both consoles and PCs, aiming to deliver a “premium and carefully crafted” gaming experience. Source
March 6, 2026
SSPAI Morning Brief: Apple Unveils MacBook Neo as OpenAI Prepares New AI Model
Morning Brief
1. Apple introduces MacBook Neo
2. Claude Code launches voice mode
3. Setapp allows individual app purchases or subscriptions
4. OpenAI confirms an upcoming new model
5. Keychron releases the B11 Pro foldable keyboard
6. ASUSTOR launches the flagship NAS AS7224RDX
Apple introduces MacBook Neo

On March 4, Apple officially unveiled the long-rumored “colorful entry-level” MacBook, with the new lineup named MacBook Neo.

Further viewing: Hands-on photos of MacBook Neo from SSPAI

MacBook Neo is powered by the A18 Pro processor, comes standard with 8GB of memory and a 13.0-inch Liquid Retina display, and features one USB 3 port and one USB 2 port. The device supports up to 16 hours of streaming video playback. It weighs 1.23 kg and includes a dual-speaker system with spatial audio support, Wi-Fi 6E, Bluetooth 6, and a 1080p FaceTime HD camera. The base model includes 256GB of storage, while the 512GB version additionally comes with Touch ID.

MacBook Neo is available in four colors: silver, peach pink, citrus yellow, and indigo blue. The 256GB version is priced at 4,599 yuan, while the 512GB version costs 5,299 yuan. The product will officially go on sale on March 11. Source

Product appearance images, provided by the news source

Claude Code launches voice mode

On March 3, Anthropic announced the gradual rollout of a new Voice Mode in its programming tool Claude Code, signaling that the developer-favored concept of “intuitive programming” is evolving toward a more efficient era of “voice programming.”

The voice module is built directly into the underlying architecture of Claude Code. After receiving beta access, users will see a prompt prominently displayed on the welcome screen inviting them to enable the feature. During actual use, developers only need to press and hold the space bar on the keyboard to speak and input development requests, then release the key to quickly send the voice command to the AI for execution.

In terms of pricing, Anthropic emphasized that the voice interaction feature will not incur any additional subscription fees. As the feature rolls out more broadly, users on the existing Pro, Max, Team, and Enterprise tiers will all be able to experience this new capability designed to improve development efficiency at no extra cost. Source

Setapp allows individual app purchases or subscriptions

On March 3, the MacPaw team announced that it will introduce new purchasing options for its Setapp platform, including standalone subscriptions and one-time purchase plans for specific apps. Under the new model, applications such as Bartender, Downie, and AlDente Pro, along with MacPaw’s own apps like CleanMyMac, Moonlock, and Gemini 2, will offer a variety of subscription and purchase plans, including monthly, yearly, and one-time purchase options. These purchase or subscription plans can all be accessed through Setapp, without requiring users to subscribe to the full Setapp service.

Currently, more than 60 apps support the new purchase and subscription options, and the list of participating applications will be updated and announced regularly in the future. Source

OpenAI confirms an upcoming new model

On March 4, shortly after launching the GPT-5.3 Instant model, the OpenAI team stated on social media that GPT-5.4 “will arrive sooner than you think.” Since the GPT client shows that the older 5.1 model will be discontinued on March 11, some users speculate that OpenAI may officially release version 5.4 around that time. Source

Keychron releases the B11 Pro foldable keyboard

On March 4, the Keychron brand introduced the B11 Pro foldable portable keyboard. The product features a flat Alice layout, which is more ergonomic than the traditional horizontal layout, while remaining compatible with macOS, Windows, and Linux. The B11 Pro adopts an inward-folding design, with an ABS body and a matte black PU leather back. It uses scissor-switch mechanisms and comes with concave ABS keycaps. In terms of connectivity, the keyboard supports tri-mode connections including USB-C wired, 2.4GHz wireless, and Bluetooth 5.3 wireless, with a “dual 1kHz” polling rate. The entire keyboard weighs 258g and includes a built-in 250mAh battery rated for up to 138 hours of battery life, supporting automatic power-off when folded and automatic power-on when unfolded. The price is set at $64.99. Source

Product appearance image, photo sourced from news outlets

ASUSTOR launches the flagship NAS AS7224RDX

On March 3, ASUSTOR, a subsidiary of ASUS, announced the launch of its 4U rackmount flagship NAS, the AS7224RDX (Lockerstor 24R PRO Gen2), featuring 24 drive bays. The AS7224RDX is powered by an AMD Ryzen 7 PRO 7745 eight-core processor and provides 24 SATA bays supporting either 3.5-inch or 2.5-inch drives. It also includes PCIe Gen5 ×8 and PCIe Gen5 ×4 expansion slots, one M.2 2280 PCIe Gen5 ×4 slot, two 10GbE network ports, and two 1GbE network ports.

In addition, the device features four DDR5 UDIMM memory slots supporting up to 192GB of ECC memory, and comes preinstalled with 16GB DDR5-4800 ECC memory and a 32GB SATA DOM. It is powered by a pair of 550W Platinum redundant power supplies and equipped with four 60mm cooling fans. The system also provides four USB-A 5Gbps ports and one HDMI port for service purposes. Source

Product appearance images, provided by the news source
March 5, 2026
SSPAI Morning Brief: Apple Unveils M5 Chips and New MacBooks as Google Rolls Out Pixel Feature Drop
Morning Brief
1. Apple unveils M5 series chips, Studio Display, and new MacBook models
2. Unihertz releases the new full-keyboard smartphone Titan 2 Elite
3. Four AI updates
4. Google releases the Spring 2026 Pixel Feature Drop
5. Speedtest and Downdetector are being sold
6. Chrome adjusts the update cycle for the stable version
7. Highguard announces shutdown on March 12
8. Briefs worth a quick look
Apple unveils M5 series chips, Studio Display, and new MacBook models

On March 3, Apple introduced the M5 series chips and simultaneously refreshed the MacBook Pro, MacBook Air, and Studio Display / Display XDR product lines.

The M5 series adopts a Fusion architecture that integrates two 3-nm chiplets into a single SoC. The M5 Pro and M5 Max feature a new 18-core CPU architecture composed of 6 super cores and 12 performance cores. Multi-thread performance improves by up to 30% compared with the M4 series and up to 2.5× compared with the M1 series. GPU configurations scale up to 40 cores, with each core integrating neural network accelerators. Peak AI compute performance improves by more than 4× over the previous generation, while ray tracing performance increases by up to 35%. In terms of memory, M5 Pro supports up to 64GB with 307GB/s bandwidth, while M5 Max supports up to 128GB with 614GB/s bandwidth. The M5 series also integrates a 16-core Neural Engine, a media engine supporting AV1 decoding, a Thunderbolt 5 controller, and memory security protection features. Source

New MacBook Pro models powered by the M5 series start at RMB 17,999 for the 14-inch M5 Pro version and RMB 21,999 for the 16-inch model; the 14-inch M5 Max version starts at RMB 29,999 and the 16-inch version at RMB 31,999. Orders open on March 4, with official sales beginning on March 11. Source

Apple also introduced 13-inch and 15-inch MacBook Air models powered by the M5 chip, featuring a 10-core CPU and up to a 10-core GPU. Base storage starts at 512GB and can be configured up to 4TB. The new generation SSD delivers twice the read and write performance. In terms of connectivity, the devices debut Apple’s N1 wireless chip, supporting Wi-Fi 7 and Bluetooth 6, and include two Thunderbolt 4 ports capable of driving up to two external displays. The machines use a fanless silent design and feature a 12MP Center Stage camera and Liquid Retina display, with battery life of up to 18 hours. The new MacBook Air is available in Sky Blue, Midnight, Starlight, and Silver, with the 13-inch model starting at RMB 8,499 and the 15-inch model at RMB 9,999. Orders open on March 4, with sales beginning March 11. Source

Finally, the new Studio Display features a 27-inch 5K Retina display with brightness up to 600 nits. The Studio Display XDR upgrades to a 27-inch 5K Retina XDR panel using mini-LED backlighting with 2,304 local dimming zones. It supports a 120Hz ProMotion adaptive refresh rate (47Hz–120Hz), SDR brightness up to 1000 nits, HDR peak brightness up to 2000 nits, a contrast ratio of 1,000,000:1, and newly adds support for the Adobe RGB color gamut. Both displays include a 12MP Center Stage camera with Desk View support, a studio-quality three-microphone array, and a six-speaker system supporting spatial audio. For connectivity, the lineup adds Thunderbolt 5 downstream support, providing two Thunderbolt 5 ports and two USB-C ports, and supports daisy-chaining up to four displays. Each unit includes a Thunderbolt 5 Pro cable. Studio Display supports up to 96W power delivery, while the XDR version supports up to 140W reverse charging.

The Studio Display with a tilt-adjustable stand starts at RMB 11,999, while the Studio Display XDR with tilt- and height-adjustable stand starts at RMB 24,999. Orders open on March 4, with sales beginning March 11. Source

Unihertz releases the new full-keyboard smartphone Titan 2 Elite

On March 3, Unihertz announced the Titan 2 Elite, a new full-keyboard smartphone in its Titan series, at MWC 2026. The device adopts a compact candy-bar design reminiscent of classic BlackBerry phones and comes in black and orange color options. It features a 4.03-inch AMOLED display with a 120Hz refresh rate, and the secondary rear display from the previous generation has been removed to achieve a slimmer and lighter design. The phone is powered by the MediaTek Dimensity 7400 processor, with an alternative Dimensity 8400 Pro version also available, and includes 12GB of RAM and 512GB of storage. The 3.5mm headphone jack has been removed, while the programmable side button and infrared transmitter remain.

On the software side, the Titan 2 Elite ships with Android 16. Unihertz promises security updates through 2031 and Android version upgrades up to Android 20. The Titan 2 Elite will launch on Kickstarter for crowdfunding in March this year, with pricing yet to be announced. Source

Four AI Updates

OpenAI Releases GPT-5.3 Instant
On March 3, OpenAI released GPT-5.3 Instant. The new model improves response relevance, conversational fluency, and web search capabilities, while significantly reducing mechanical replies, redundant disclaimers, and overly stiff tone tendencies. Compared with GPT-5.2 Instant, it particularly improves responses involving sensitive topics, addressing previous issues where the model could appear overly cautious or overly preachy. It also enhances the integration quality of web search results, reduces excessive reliance on search outputs, and resolves earlier cases where responses occasionally contained “link dumping” or loosely organized information. The update also substantially adjusts dialogue styles that users previously described as “awkward,” “out of place,” or overly forceful in tone, while significantly reducing hallucination rates across a wide range of topics.

GPT-5.3 Instant is available immediately to all ChatGPT users. Developers can access the model through the API under the name gpt-5.3-chat-latest. Updates for the Thinking and Pro variants are expected to roll out soon. GPT-5.2 Instant will remain available for paid users under the “Legacy Models” dropdown menu for three months and is scheduled to be officially retired on June 3, 2026. Source

Google Releases Gemini 3.1 Flash-Lite
On March 3, Google officially launched Gemini 3.1 Flash-Lite. The model features a context window of up to one million tokens and supports multimodal inputs including text, images, audio, and video. It delivers a maximum output speed of 389 tokens per second. Pricing is set at $0.25 per million input tokens and $1.50 per million output tokens. The model’s knowledge cutoff date is January 1, 2025.

A preview version of Gemini 3.1 Flash-Lite is now available to developers via Google AI Studio and the Vertex AI platform. Source

Anthropic Introduces Memory Import Tool
On March 3, Anthropic announced that the memory feature of its AI assistant Claude is now available to free users. The feature is implemented through standardized prompts: users can copy a specific prompt provided by Anthropic into other AI chat tools and request them to export all stored user preferences, personal details, project goals, and behavioral adjustments. The returned structured text can then be pasted into Claude’s memory settings to complete the import and integration process. Source

Alibaba Releases Qwen3.5 Small Model Series
On March 2, Alibaba’s Qwen team released the Qwen3.5 Small model series, including four parameter sizes: 0.8B, 2B, 4B, and 9B. Built on the Qwen3.5 architecture, the series features native multimodal capabilities and has been optimized through large-scale reinforcement learning (RL). The 0.8B and 2B versions focus on high-speed inference for edge devices, the 4B version is positioned as a lightweight multimodal foundation for AI agents, and the 9B version aims to narrow the performance gap with larger models. Base models are also provided across the entire lineup. The weights have been released on Hugging Face and ModelScope, where they are freely available for research and industrial applications. Source

Google releases the Spring 2026 Pixel Feature Drop

On March 3, Google rolled out the March Pixel Drop for Pixel devices. The update adds real-time location sharing in Google Messages to the Find Hub tracking network and introduces lost luggage tracking in partnership with 10 airlines (requiring compatible tags). A customizable Calling Card feature is also introduced, allowing users to personalize their call screens with photos and fonts, while the Google Play Store adds a short-video feed to enhance app discovery. Pixel 10 series–exclusive Magic Cue now supports restaurant recommendations, and the Circle to Search feature expands with enhanced virtual try-on and video shopping capabilities. The home screen also gains six AI-generated icon styles. Pixel Watch receives several updates, including phone-loss alerts, quick payments without waking the device, earthquake alerts, and satellite SOS coverage across Europe, Canada, Hawaii, and Alaska. These updates will roll out gradually starting today. Source

Speedtest and Downdetector are being sold

On March 3, digital media company Ziff Davis announced that it will sell its Connectivity business unit, which includes Ookla Speedtest and Downdetector, to Accenture for $1.2 billion in cash. The division generated $231 million in revenue in 2025, representing a substantial return compared with the $15 million Ziff Davis paid to acquire it in 2014. Ziff Davis stated that the deal is intended to sharpen the company’s strategic focus, with future efforts centered on core media brands such as IGN, Mashable, and Everyday Health. The transaction is expected to close in the coming months, and both Speedtest and Downdetector will continue operating normally during the transition period. Source

Chrome adjusts the update cycle for the stable version

On March 3, Google announced that the update frequency for the Chrome browser’s stable release across all platforms will change from once every four weeks to once every two weeks. The adjustment applies to desktop, Android, and iOS platforms, aiming to accelerate feature delivery and simplify post-release debugging. The first stable version under the new schedule will be Chrome 153, scheduled for release on September 8. Beta versions will be released three weeks ahead of the stable version, while weekly security updates for milestone versions will remain unchanged. The Canary and Dev channels, as well as the extended eight-week stable release cycle for enterprise users, will continue as before.

For Chromebook hardware, ChromeOS will follow the new cadence after completing platform-specific testing. Further details regarding update policies for managed devices will be announced later. Source

Highguard announces shutdown on March 12

On March 3, developer Wildlight Entertainment announced that its shooter game Highguard will officially shut down on March 12. The title was developed by a team composed of former members from Apex Legends, Call of Duty, and Titanfall, and was showcased as the closing highlight at The Game Awards 2025. Since its release on January 26, 2026, the game once attracted up to 2 million players but ultimately failed to establish a player base large enough to sustain long-term operation. The studio is reportedly partially funded by Tencent and laid off most of its staff in February. The game will receive its final update over the next two days, including a new character and new weapons. Source

Briefs worth a quick look
- Swedish media outlet Svenska Dagbladet reported that Meta AI smart glasses (such as Ray-Ban Meta) allegedly captured users’ private videos and financial information without their knowledge and sent the data to reviewers outside the European Union. The claim was revealed by an AI data annotator in Nairobi, Kenya, who said that while processing visual data used to train large language models (LLMs), they encountered highly sensitive content including nudity, people using the toilet, intimate behavior, and even credit card numbers. Meta declined to comment on the report. Source
- Apple’s official regulatory compliance page briefly listed a previously unknown device called MacBook Neo (Model A3404), identified as part of the 2026 MacBook lineup, though the page has since been removed. According to earlier rumors, the device is expected to target the education and consumer entry-level market and may use an iPhone-series processor rather than an M-series chip. Source
March 4, 2026
SSPAI Morning Brief: Apple Unveils iPhone 17e and M4 iPad Air as Qualcomm Introduces Snapdragon Wear Elite
Morning Brief
1. Apple unveils iPhone 17e and M4-powered iPad Air
2. Two updates from Motorola
3. Lenovo announces multiple new products at MWC
4. HONOR unveils several new products at MWC
5. Qualcomm introduces the new Snapdragon Wear Elite
6. iReader launches the iReader Tango 2 Pro e-ink reader
7. Three updates from Alibaba’s Qwen
8. Anker eufyMake E1 3D Texture UV Printer now available domestically
9. Briefs worth a quick look
Apple unveils iPhone 17e and M4-powered iPad Air

On March 2, Apple announced two new products: the iPhone 17e and a new iPad Air powered by the M4 chip.

The iPhone 17e features a 6.1-inch display with Ceramic Shield 2 and an aluminum frame. It is powered by the A19 chip with a 4-core GPU and the C1X modem, and supports eSIM. Compared with the C1 modem found in the iPhone 16e, the C1X delivers up to twice the speed. The device supports MagSafe, with wireless charging up to 15W, and offers up to 26 hours of video playback. The iPhone 17e starts at 256GB of storage and is available in black, white, and pink, with pricing starting at RMB 4,499. It will go on sale on March 11. Source

The new iPad Air is equipped with the M4 chip (8-core CPU and 9-core GPU), with unified memory increased to 12GB. It is available in 11-inch and 13-inch sizes, and comes in Space Gray, Blue, Purple, and Starlight. Storage starts at 128GB, with pricing beginning at RMB 4,799. It will also go on sale on March 11. The new iPad Air adopts the N1 chip and C1X modem, and supports Wi-Fi 7, Bluetooth 6, and Thread connectivity. Source

Two updates from Motorola

Motorola Partners with GrapheneOS

On March 2 at MWC 2026, Motorola announced a long-term partnership with GrapheneOS, aiming to bring advanced security technologies to everyday users. Through this collaboration, the two parties will work together to enhance smartphone security and explore future device compatibility with GrapheneOS. Motorola will collaborate with GrapheneOS in the coming months on software features, security enhancements, and related areas, with more details to be revealed as the partnership progresses. At present, Motorola has not disclosed which models will be compatible with or ship with GrapheneOS. Previously, GrapheneOS was only available on Google Pixel devices. Source

Motorola Unveils the Razr Fold Horizontal Foldable

At MWC 2026, Motorola introduced its first horizontally folding smartphone, the Razr Fold. The device is powered by the Snapdragon 8 Gen 5 processor, paired with 16GB of RAM and 512GB of storage, with a 1TB variant available in select markets. It features a 6.6-inch 165Hz 21:9 pOLED external display, making it the first smartphone to use Corning Gorilla Glass Ceramic 3. The internal display is an 8.1-inch 120Hz LTPO panel. The device measures 4.6mm thick when unfolded (excluding the camera module), 9.9mm when folded, and weighs 244g. The Razr Fold packs a 6,000mAh battery, supporting 80W TurboPower fast charging, 50W wireless charging, and 5W reverse wireless charging. It carries IP48 and IP49 ratings for dust and water resistance, and supports the Moto Pen Ultra stylus. The Razr Fold will ship with Android 16 and is promised at least seven major OS upgrades along with seven years of security updates. It will be available in Pantone Blackened Blue and Pantone Lily White. Pricing and exact availability have yet to be announced, but the device is set for release within 2026. Source

Lenovo announces multiple new products at MWC

At MWC 2026, Lenovo introduced a range of new products and concept devices.

The Yoga Book Pro 3D concept laptop highlights its glasses-free 3D display, aiming to naturally present depth, shape, and spatial relationships on screen. The laptop features dual Lenovo PureSight Pro Tandem OLED displays. Powered by Lenovo’s in-house AI software, 2D images can be converted into 3D and shown on the upper display. Users can also use mid-air gestures captured via an RGB camera to zoom, rotate, and interact with content. The lower display supports an attachable keyboard, offering shortcuts for adjusting lighting, viewing angles, and color tones. The device is powered by an Intel Core Ultra 7 processor and an RTX 5070 GPU.

The Legion Go Fold concept handheld features a POLED display that expands from 7.7 inches to 11.6 inches, with detachable controllers. It supports four usage modes: standard handheld mode, vertical split-screen mode, horizontal full-screen mode, and extended desktop mode.

In extended desktop mode, the right controller can function as a vertical mouse. It includes a small display that can act as a touchpad, show game settings and performance metrics, or serve as customizable hotkeys. The Legion Go Fold is powered by an Intel Core Ultra 7 258V processor, features a 48Whr battery, and comes with 32GB of RAM.

The Yoga 9i 2-in-1 Aura Edition (14″, 11) retains the Yoga series’ classic 360° convertible design. Lenovo positions it as a versatile drawing tablet. It features a 2.8K PureSight Pro OLED touchscreen with a peak brightness of 1100 nits and 100% coverage of DCI-P3, Adobe RGB, and sRGB color gamuts. It supports the Yoga Pen Gen 2 stylus, and the pen case magnetically attaches to the lid to create a canvas mode for improved ergonomics while drawing.

The device continues to support the Yoga series’ multi-mode usage. It is powered by an Intel Core Ultra Series 3 processor and qualifies as a Copilot+ PC. Ports include two USB-C Thunderbolt 4 ports, one USB-A 3.2 Gen 2 port, and one HDMI 2.1 FRL port. It weighs 1.29kg and is expected to launch in March this year.

The Yoga Pro 7a (15″, 11) is equipped with AMD Ryzen AI Max+ series processors and supports a unified memory architecture, allowing the CPU and GPU to share up to 128GB of RAM. It is compatible with the Yoga Pen Gen 2, and its Force Pad touchpad is powered by Wacom technology for drawing, automatically disabling touch gestures when the stylus is in use. The device features Lenovo X Power thermal technology, upgraded heat pipes, and dual fans, increasing thermal performance to 95W TDP while keeping noise levels as low as 22dB. It is expected to launch in June this year.

The IdeaPad Pro Gen 2 tablet runs on the Snapdragon 8s Gen 4 mobile platform and integrates Lenovo AI solutions. It features a 13-inch 3.5K PureSight Pro display with Dolby Vision support. The tablet measures 6.2mm thick, weighs under 600g, includes a 10,200mAh battery with 45W fast charging, and is available in Moon Gray, Cloud Gray, and Mint.

The IdeaPad Slim 5i Ultra (14″, 11) laptop offers either a WUXGA OLED or WQXGA IPS LCD 120Hz display. It houses a 65Wh battery, supporting up to approximately 20 hours of video playback. The device weighs as little as 1.15kg and measures 11.9mm at its thinnest point. It includes an FHD infrared camera for facial recognition and features a physical privacy shutter.

The Legion Tab (8.8″, 5), known domestically as the Y700 (5th Gen), features an 8.8-inch 3K display with a 165Hz refresh rate and typical brightness of 600 nits. It is powered by the Snapdragon 8 Elite Gen 5 mobile platform, with configurations up to 16GB RAM and 512GB storage. The 9,000mAh battery is paired with Lenovo’s Legion Coldfront Vapor cooling system, improving thermal efficiency by 32% compared to the previous generation. The tablet weighs 360g and comes in black, white, and fluorescent green.

The Legion 7a (15″, 11) gaming laptop also features AMD Ryzen AI Max+ series processors with unified memory architecture. It uses an aluminum chassis, weighs 1.65kg, and sports a 15.3-inch PureSight OLED display.

Lenovo also showcased an AI Work Companion concept device resembling a desktop clock. Designed for interactive engagement, it uses AI to sync daily tasks and schedules across the user’s devices and can generate a balanced daily plan with a single tap. The device includes a built-in hub for desktop charging and features customizable physical buttons that can be configured as needed.

Also announced were the Yoga Slim 7i Aura Edition, IdeaPad Slim 3i, L16 desktop monitor, and other new devices. Source

HONOR unveils several new products at MWC

At MWC 2026, HONOR announced several new products. The foldable flagship HONOR Magic V6 debuted in a new “Chitu Red” colorway, with its official domestic launch scheduled for March 10. HONOR also announced a strategic technology partnership with ARRI, aiming to “bring cinematic storytelling experiences to creators worldwide.”

HONOR introduced the Robot Phone concept device as a new form of AI terminal. The Robot Phone features a motorized four-joint movable gimbal built into its camera module. The camera can perform AI-driven interactions and real-time feedback through movement, while also supporting three-axis image stabilization and intelligent object tracking. It can even “dance” in sync with music for interactive experiences. HONOR has launched a Robot Phone wish program, allowing users to receive updates on development progress and participate in user research. The device is envisioned to launch within 2026. Source

At MWC 2026, HONOR also unveiled its first humanoid robot, HONOR ROBOT, which took part in a live dance performance, showcasing moves such as the moonwalk and a backflip. Source

Qualcomm introduces the new Snapdragon Wear Elite

On March 2 at MWC 2026, Qualcomm officially unveiled the new Snapdragon Wear Elite platform. The platform is designed to enable OEMs and AI cloud service providers to develop and innovate rapidly in this emerging space, deploying AI agents across a wide range of device form factors, including smartwatches, pin-style wearables, pendants, and more. The platform places significant emphasis on its integrated NPU, which has been substantially enhanced to run models with up to 2 billion parameters directly on-device. Combined with multimodal inputs from voice, vision, location, and various sensors, it enables highly personalized AI agents to support users in work, study, health, and everyday life. In terms of performance, the platform adopts a new five-core CPU architecture and integrates upgraded CPU and GPU components, delivering substantial performance gains. Compared to the previous generation, CPU performance improves by up to five times, while GPU performance increases by up to seven times. On the efficiency front, Qualcomm states that the platform can extend daily operating usage (DOU) by 30%, and can charge to approximately 50% in just 10 minutes. For connectivity, Qualcomm introduces what it describes as the most comprehensive connectivity suite yet for a wearable platform, supporting six different technologies: 5G RedCap, ultra-low-power Wi-Fi, Bluetooth 6.0, UWB (Ultra Wideband), GNSS (Global Navigation Satellite System), and NB-NTN satellite communication. Source

iReader launches the iReader Tango 2 Pro e-ink reader

On March 2, iReader unveiled the iReader Tango 2 Pro e-ink reader. The device adopts a smartphone-like form factor and is available in black and white. It features a newly developed in-house ultra-refresh technology, delivering a dynamic refresh rate of up to 50 frames per second, with flicker reduced to nearly zero. The reader supports SIM card insertion for cellular connectivity and enables WeChat voice functionality. An 8MP rear camera is included for QR code scanning. It uses a Carta 1300 e-ink display with freely adjustable color temperature and brightness. Powered by an 8-core CPU built on a 6nm process, the device comes with 8GB RAM and 128GB storage, with rated battery life of approximately 101 hours. The iReader Tango 2 Pro will go on sale on March 10, with an introductory price of RMB 1,999. Source

Three updates from Alibaba’s Qwen

Alibaba Unifies Its Large Model Brand Under Qwen
On March 2, Alibaba’s large model team officially announced that it has unified the overall name and core brand of Alibaba’s AI under “Qwen.” The Qwen large model lineup covers both foundational models and domain-specific models, while the Qwen App serves as Alibaba’s flagship AI application for consumers. Following the rebranding, the Chinese name of Alibaba’s large model brand is 千问大模型, with the English name Qwen. “Tongyi Lab” will remain the organizational name of Alibaba Group’s AI research institution. Source

Tongyi Lab Releases Two Natural Language–Controlled Speech Generation Models
On March 2, Alibaba Cloud’s Tongyi Lab announced the release of two speech generation models built on the new FreeStyle paradigm: Fun-CosyVoice 3.5 and Fun-AudioGen-VD, aiming to push speech generation into a new stage of fine-grained control via natural language. Fun-CosyVoice 3.5 focuses on voice cloning and expressive optimization, with its core breakthrough being the ability to directly control generation results through natural language instructions such as “make the tone more firm,” without relying on preset tags. It supports multilingual voice cloning and adds new generation capabilities for Thai, Indonesian, Portuguese, and Vietnamese, while significantly improving pronunciation accuracy, prosody quality, and response latency. Fun-AudioGen-VD, on the other hand, focuses on building voices and auditory scenes from scratch. Based on text descriptions, it can directly generate specific timbres, emotions, characters, and even complete background environments and spatial acoustic effects, enabling integrated “character + scene” audio creation. Source

Qwen AI Glasses G1 Open for Reservations Across All Channels
Qwen’s first AI hardware product, the Qwen AI Glasses, went online for reservations on March 2 and will go on sale in China on March 8. The initial launch includes two series, S1 and G1, both powered by the latest generation Qwen large model. The devices offer industry-leading response speed and interaction smoothness, support high-precision multimodal understanding and real-time interaction, and cover key daily and office scenarios including high-definition photography, AI translation, meeting transcription, and image recognition. The G1 series, which will be available first, features dual flagship chips with dual systems, a five-microphone array with bone conduction, and high-performance large-diaphragm speakers. It adopts a lightweight design with a swappable battery experience, weighing only about 40g—comparable to ordinary optical glasses—and is equipped with an adjustable swan-neck bracket and FDA food-grade silicone nose pads for all-day comfort. The G1 integrates “smartphone-level” imaging technology, enabling 0.6-second rapid capture and 3K video recording, with AI super-resolution and frame interpolation to output 4K video. It also uses Super RAW low-light enhancement technology to significantly improve image quality and stability in low-light environments. With 64GB of built-in storage, it supports extended continuous shooting. The G1 series starts at RMB 2,899, with a subsidized price of RMB 1,997 after national incentives. Source

Anker eufyMake E1 3D Texture UV Printer now available domestically

On March 2, Anker announced via its official WeChat account that the eufyMake E1 is now exclusively available on JD.com, with full-payment pre-orders set to begin on March 12. Designed for makers and small studios, the printer supports 3D textured printing on a variety of materials. It can print on metal, glass, acrylic, and wood surfaces, simulating textures such as oil painting brushstrokes and embossed leather, with a maximum print height of up to 5mm. The companion app includes AI-powered creative features, allowing users to generate textures and relief effects. The printer supports flat printing, cylindrical printing, and UV transfer stickers. It also features a built-in JetClean self-cleaning system that automatically maintains the printheads to reduce the risk of clogging. Source

Briefs worth a quick look
- At MWC 2026, Leica joined Xiaomi in unveiling the Leica Leitzphone. In China, the device will be sold as the Xiaomi 17 Ultra Leica Edition in a black and silver colorway, with pricing starting at RMB 7,999. Source
- Xiaomi has launched the Xiaomi Tag tracker, available in BLE and UWB versions, priced at RMB 69 and RMB 129 respectively. The Xiaomi Tag supports IP67 dust and water resistance and is powered by a CR2032 coin-cell battery. It is compatible with the iOS Find My app. The UWB version works with the Xiaomi 17 Ultra, Xiaomi 17 Pro Max, and Xiaomi 17 Pro. The BLE version is already on sale. Source
March 3, 2026
SSPAI Morning Brief: OpenAI Partners with Pentagon as NVIDIA Prepares AI Inference Chips
Morning Brief
1. OpenAI replaces Anthropic to partner with the Pentagon
2. Paramount set to acquire Warner Bros., Netflix out
3. South Korea lifts data restrictions on Google Maps
4. Ultrahuman launches Ring Pro
5. NVIDIA to introduce AI inference-focused chips
6. Rising memory costs may eliminate entry-level PCs
7. Rumors Worth a Quick Look
OpenAI replaces Anthropic to partner with the Pentagon

According to reports from The Wall Street Journal and other outlets, on February 27, the Trump administration announced a comprehensive ban on federal agencies using technology from AI startup Anthropic, designating it as a “supply chain risk.” At the same time, the U.S. Department of Defense reached an agreement with Anthropic’s competitor OpenAI, allowing its AI models to be deployed in classified military environments.

Previously, the Pentagon had requested absolute usage rights for the Claude model within all legal boundaries, while Anthropic insisted on maintaining red lines prohibiting use in domestic mass surveillance and autonomous weapons. After negotiations broke down, Trump accused Anthropic of being a “radical left-wing” company on social media and ordered all federal agencies to fully discontinue use of its models within six months. The U.S. Department of Defense subsequently placed the company on a “supply chain risk” list typically reserved for foreign adversarial firms. This move not only cut off approximately $200 million in government contracts but may also compel other U.S. defense contractors to divest from the technology. Anthropic has stated it will challenge the decision in court.

Following Anthropic’s exit, OpenAI quickly filled the gap in military contracts. Sam Altman stated that the new agreement with the military includes safety provisions that similarly prohibit use for mass surveillance or autonomous weapons. OpenAI published the specific wording of these clauses on its website, noting that it is unclear why Anthropic was unable to reach a similarly secure agreement.

OpenAI emphasized that the partnered models remain cloud-hosted and therefore retain autonomous controllability — a solution that Anthropic had previously evaluated and explicitly rejected during negotiations, arguing that modern military systems are already highly networked, rendering any distinction between cloud and weapon endpoints largely meaningless. Additionally, OpenAI’s terms prohibit AI from directly controlling weapons only in scenarios where laws or departmental policies require human oversight. However, current Department of Defense policy relies on vague language such as requiring an “appropriate level of human judgment,” which could potentially be modified to bypass such restrictions. Safeguards related to mass surveillance largely restate existing applicable laws and do not prohibit governments from obtaining surveillance data indirectly or through third parties, whereas Anthropic had sought to explicitly bar the government from using AI to analyze “non-classified bulk commercial data.”

Media reports indicate that OpenAI executives donated tens of millions of dollars last year to political action committees supporting Trump, while Anthropic’s CEO reportedly had personal conflicts with the Department of Defense’s CTO. By the day negotiations concluded, Trump had already prepared social media posts criticizing Anthropic. While Anthropic representatives were still attempting to continue talks, a replacement agreement between the Department of Defense and OpenAI was already in place.

Founded in 2021 by several core researchers who left OpenAI due to concerns over AI safety, Anthropic is currently among the most highly valued AI companies in the world. The company has maintained a strained relationship with the current U.S. government, having drawn criticism for hiring former Biden administration officials, advocating for stronger AI regulation, and opposing restrictions on AI chip exports to China. In January this year, the U.S. military reportedly used the Claude model in operations in Venezuela. Anthropic subsequently inquired with partners about the model’s usage details, which the Pentagon viewed as a challenge to military decision-making authority — an incident that ultimately contributed to the breakdown in relations. Just hours before negotiations collapsed, Claude was reportedly used in a U.S. airstrike operation in the Middle East.

The incident has sparked a wave of U.S. users uninstalling OpenAI’s ChatGPT in favor of Claude. By Saturday, Claude had risen to the top of the U.S. Apple App Store’s free app rankings, surpassing ChatGPT.

Paramount set to acquire Warner Bros., Netflix out

According to The Wall Street Journal, on the evening of February 26, after six months of intense bidding, Paramount — led by David Ellison — successfully acquired Warner Bros. Discovery in a deal valued at up to $81 billion. Streaming giant Netflix officially withdrew from the bidding, abandoning its previously agreed $72 billion acquisition proposal with Warner.

David Ellison is the son of Oracle founder Larry Ellison. Prior to this, he had secured control of Paramount through his company Skydance Media. Following the acquisition of Warner, Ellison will preside over an unprecedented media empire, bringing HBO, CNN, CBS News, Warner Bros. Pictures, as well as major IP such as DC Comics and Harry Potter, under his control.

The bidding battle proved exceptionally complex, with Ellison submitting nine separate offers over the course of six months. Warner initially rejected multiple proposals and even announced in December last year that it would accept Netflix’s acquisition plan. However, Ellison refused to back down, raising his bid to $31 per share — a 63% increase over his initial offer — while pledging to cover Warner’s $2.8 billion termination fee to Netflix. He also threatened to challenge control at the shareholder level and engaged in extensive political lobbying in Washington, ultimately persuading Warner’s board to reverse its decision.

The news triggered a strong reaction in the capital markets, with Paramount’s stock surging more than 20% on Friday, while Netflix shares — having exited the bidding war — also rose nearly 14%. The transaction is expected to take between 6 to 18 months to close. The merged Paramount is projected to carry approximately $100 billion in debt, with management facing the challenge of cutting $6 billion in costs.

South Korea lifts data restrictions on Google Maps

According to The New York Times, on February 27, South Korea’s Ministry of Land, Infrastructure and Transport approved Google’s request to export high-precision map data from the country. This decision overturns longstanding regulatory restrictions and clears the way for Google Maps to offer full functionality in South Korea, including real-time navigation.

For years, citing national security concerns, the South Korean government strictly limited the overseas transfer of mapping data. As a result, Google Maps has been unable to provide real-time navigation, walking directions, or detailed business listings within the country, forcing both residents and foreign visitors to rely on domestic navigation apps. Under the newly approved conditions, Google will be permitted to transfer government-reviewed map data to its global data centers, provided that sensitive security information such as military base locations is strictly excluded. Google executives welcomed the decision but have yet to disclose a timeline for the rollout of full navigation features.

Since introducing its map services in South Korea in 2008, Google has faced ongoing limitations due to these policies, and first publicly requested permission to export mapping data in 2016. According to the Ministry, the policy shift is intended to “boost tourism and stimulate economic growth,” though it has also been influenced by U.S.–South Korea trade negotiations. Previously, the Trump administration had reportedly pressured Seoul to cease regulatory practices that disadvantaged U.S. technology firms in areas such as location data.

The full deployment of Google Maps is expected to directly challenge the dominance of domestic players such as Naver and Kakao in South Korea’s digital mapping sector. The country’s spatial information industry has expressed concern that the entry of a global platform could ultimately reduce local firms to subcontractors for Google. However, some tourism and business stakeholders remain optimistic, believing that the move will help local merchants expand into international markets while encouraging domestic mapping services to improve through healthy competition.

Ultrahuman launches Ring Pro

According to The Verge, on February 27, health tech brand Ultrahuman unveiled its third-generation flagship smart ring, Ring Pro, along with the AI health platform Jade. Ring Pro is available in four color options and is priced at $479. It is currently open for global pre-orders and will begin shipping in March. However, due to an ongoing patent dispute with leading smart ring brand Oura, the new device is not yet available in the U.S. market.

In terms of hardware, Ring Pro’s battery life has been extended from the previous 4 to 6 days to a maximum of 15 days on a single charge, and it now supports offline storage of up to 250 days of health data. The device also features a faster dual-core processor, an upgraded heart rate sensor, and safety-focused physical design improvements to ensure it can be more easily removed from the finger in emergency situations.

Also introduced is a new Pro charging case, capable of providing an additional 45 days of battery life while storing up to one year of device data. The charging case supports wireless charging, includes an LED battery indicator, and comes equipped with a built-in speaker and proximity tracker.

The Jade bio-intelligent AI platform is currently capable of delivering real-time health insights, guiding breathing exercises, and detecting atrial fibrillation (AFib). Ultrahuman plans to expand its capabilities through future updates, enabling proactive health risk alerts, smart home temperature adjustments, and even online food ordering — with the goal of transforming Jade into an always-on “autonomous health agent.”

NVIDIA to introduce AI inference-focused chips

According to The Wall Street Journal, NVIDIA plans to unveil a new chip specifically designed for AI inference — the process by which models respond to user prompts — at its upcoming GTC Developer Conference in San Jose next month. The product integrates chip designs from startup Groq, enabling more efficient handling of complex AI requests. OpenAI, the developer of ChatGPT, has agreed to become one of the largest customers for the new chip.

As enterprises increasingly deploy AI agents capable of autonomously executing tasks and automated programming tools, demand for inference computing power has surged. Although NVIDIA currently holds over 90% of the global GPU market share with its Hopper and Blackwell series, traditional GPUs remain costly and energy-intensive when handling real-world inference workloads.

To address competition from in-house chips developed by Google and Amazon, as well as startups such as Cerebras, NVIDIA invested heavily late last year to bring Groq’s technology and core team under its umbrella. Groq utilizes a distinct architecture known as a Language Processing Unit (LPU), which delivers exceptional efficiency during the token-by-token decoding phase of AI model response generation. OpenAI is reportedly planning to leverage NVIDIA’s new system to enhance its automated programming tool Codex in response to competing solutions such as Claude Code, which primarily run on Amazon and Google cloud chips.

In an effort to retain key clients, NVIDIA recently invested $30 billion in OpenAI and secured substantial orders for dedicated inference compute capacity, offsetting OpenAI’s recent procurement of chips from Amazon and Cerebras. Additionally, NVIDIA expanded its partnership with Meta this month, deploying large-scale server clusters powered exclusively by its self-developed Vera CPUs for the first time, supporting Meta’s AI agents for advertising-related computational workloads.

Rising memory costs may eliminate entry-level PCs

According to The Register, citing forecasts from research firm Gartner, surging memory demand driven by the AI boom is creating a severe cost crisis for the global PC and smartphone markets. As DRAM and NAND flash prices continue to climb, global PC shipments may decline by more than 10% this year, while smartphone shipments are expected to drop by around 8%. Entry-level devices priced below $500 may disappear from the market entirely.

Unlike previous price hikes caused by production cutbacks, the current surge is largely driven by massive demand from hyperscale cloud providers for AI computing infrastructure, directly pushing up component costs across the consumer electronics sector. Some memory prices have already multiplied several times since last year, and Gartner predicts that the cost of memory and storage used in personal devices could rise by a further 130% by the end of 2026. Under this pressure, device manufacturers are finding it increasingly difficult to maintain profit margins on budget PCs, while raising prices risks alienating price-sensitive consumers.

The adoption of AI PCs is also being hindered by elevated memory costs. For instance, Microsoft’s Copilot+ platform requires a minimum of 16GB of memory, significantly raising manufacturing thresholds. HP recently revealed that DRAM now accounts for 35% of total PC bill-of-materials costs. As a result, AI PCs are likely to remain positioned in the premium segment in the near term, with market share not expected to exceed 50% until 2028. The smartphone sector faces similar challenges, with low-end models being hit particularly hard due to the higher proportion of storage components in their overall cost structure.

As the cost of new devices rises across the board, both enterprises and consumers will be forced to extend the lifespan of existing hardware. Replacement cycles for commercial and consumer PCs are projected to lengthen by 15% and 20%, respectively. Analysts warn that this demand-driven memory shortage may persist for the long term, potentially lasting through the end of 2027.

Rumors Worth a Quick Look
- On February 28, Google announced that it will automatically restore developer accounts for Gemini CLI and Antigravity that were recently suspended for violating service terms, and has introduced a self-service account recovery process. Affected accounts are expected to regain access within one to two days. Going forward, Google will no longer impose unannounced bans. Users flagged for violations will receive email notifications with clear error messages and a dedicated appeal form link. To restore access, users must reaffirm compliance with the terms of service, including a commitment not to use third-party tools to bypass system restrictions. Once submitted, accounts will be automatically reinstated through periodic synchronization. However, a second violation will result in permanent suspension. Google emphasized that any attempt to use third-party software, tools, or services (such as OpenClaw) to “harvest” or “piggyback” Gemini CLI’s OAuth authentication in order to access Google backend services or usage quotas constitutes a serious violation.
- On February 26, Restaurant Brands International (RBI), the parent company of Burger King, announced plans to roll out an AI assistant named Patty across its U.S. locations. The system is designed to support employees and monitor customer interactions in order to assess overall team friendliness. Addressing concerns over workplace surveillance, Burger King stated that the AI is not intended to monitor or evaluate individual staff, nor will it require employees to follow scripted responses. Instead, it aggregates the frequency of keywords such as “welcome,” “please,” and “thank you” to help managers understand overall service patterns. Previously, brands such as McDonald’s and Taco Bell experimented with AI-powered drive-through ordering systems, but these were discontinued due to frequent errors.
- According to The Register, Microsoft’s HoloLens has found a new role within the U.S. military. Initially tested in 2018 as battlefield equipment, the device was shelved after causing severe physical discomfort among soldiers, including headaches, visual fatigue, and nausea. It is now being repurposed by the U.S. Army and Air Force to assist with remote aerial cargo inspections, with early success reported. In practice, Army personnel responsible for loading supplies wear the HoloLens headset, while Air Force experts connect remotely via laptops and Wi-Fi hotspots. Using a first-person perspective and AR visual prompts, Air Force staff guide soldiers in adjusting rigging and repositioning cargo to ensure proper weight distribution for safe airdrops. According to military officials, the system has reached a “plug-and-play” level of readiness after a year of refinement. Microsoft ended further development of HoloLens in late 2024, with official support for existing models set to conclude by the end of 2027. It remains unclear whether the headsets used in these trials are drawn from military inventory or newly procured.
- On March 1, Xingji Meizu Group’s China CMO Wan Zhiqiang stated during a livestream on Meizu’s JD flagship store that existing Meizu models will continue to be sold at current prices, though some are already sold out or nearing depletion. The Meizu 23 has completed development but will not be released commercially. Future updates to the Flyme operating system will be limited to essential maintenance, with no major version upgrades planned. Previously, on February 27, Meizu Technology announced a strategic transformation, stating that it will suspend domestic self-developed hardware projects for new smartphone products.
- Mark Gurman reports that —
- Apple plans to release a touchscreen MacBook Pro by the end of 2026, while maintaining a clear distinction between Mac and iPad product lines. The new MacBook Pro will retain its traditional design and macOS interface, with touch input offered as an optional feature rather than a primary interaction method. A large foldable iPad is also reportedly in development for release later in the decade, and will continue to run iPadOS. A redesigned MacBook Air is not expected until around 2028.
- At this year’s WWDC, Apple is also expected to rebrand the Core ML framework as Core AI.
March 2, 2026