Category: NEWS

Morning Brief

1. Microsoft announces improvements and simplification to the Windows Insider Program
2. Hong Kong issues stablecoin issuer licenses to HSBC and Standard Chartered
3. Red Hat lays off its China R&D team
4. The Linux kernel project introduces rules for AI-generated code submissions
5. Betting on weather becomes a trending trading strategy
6. Cyberspace Administration and Railway Authority summon third-party train ticket platforms for talks
7. News Worth a Quick Look

Microsoft announces improvements and simplification to the Windows Insider Program

On April 10, Microsoft announced improvements to the Windows Insider Program to address long-standing complaints about its increasingly fragmented structure.

The revamped program will streamline multiple channels into two: an Experimental channel and a Beta channel. The Experimental channel replaces the former Dev and Canary channels, allowing users to try cutting-edge features still under active development; the new Beta channel will provide features a few weeks ahead of their release to the stable version. The existing Release Preview channel will be retained but moved under advanced options, primarily serving enterprise customers who need early access to near-final builds.

The feature rollout mechanism is also being adjusted. In the past, Microsoft implemented a “controlled feature rollout” system in the name of quality assurance, meaning users within the same channel often received new features at different times. Going forward, the Beta channel will completely eliminate this gradual rollout approach, allowing users to access all officially announced features immediately after updating. Meanwhile, Microsoft will introduce a Feature Flags page in the Experimental channel settings, enabling advanced users to manually toggle specific features on or off.

In addition, users previously had to reinstall their systems if they wanted to switch channels or exit the Insider Program entirely. To lower this barrier, Microsoft will introduce an in-place upgrade mechanism. Except in rare cases—such as when running builds based on future system foundations—users will be able to switch between channels or exit the program without losing apps, settings, or personal data.

Hong Kong issues stablecoin issuer licenses to HSBC and Standard Chartered

According to Caixin, on April 10, the Hong Kong Monetary Authority (HKMA) announced the issuance of its first batch of stablecoin licenses to RD Innotech Limited and HSBC. RD Innotech is a joint venture formed by Standard Chartered, HKT, and Animoca Brands. Stablecoins are cryptocurrencies backed by fiat currencies, commodities, or other assets, meaning their value is not entirely determined by market forces and typically does not deviate significantly from their underlying peg.

The HKMA stated that both licensed issuers plan to launch Hong Kong dollar-denominated stablecoins in the initial phase, targeting four key application areas: cross-border payments, local payments, tokenized asset trading, and supply chain financing. RD Innotech is expected to roll out HKDAP in phases starting in the second quarter of 2026, while HSBC plans to launch its HKD stablecoin in the second half of 2026, integrating it with widely used services such as PayMe and the HSBC Hong Kong mobile banking app.

A deputy chief executive of the HKMA noted that the choice of currency is determined by the issuer’s business plan rather than regulatory requirements. If issuers wish to launch stablecoins denominated in other currencies, they must submit proposals for approval, which will be evaluated alongside regulatory requirements in other jurisdictions.

The HKMA reported receiving 36 applications, with evaluation criteria focusing on applicants’ risk management capabilities, regulatory compliance across jurisdictions, and the feasibility of their proposed business models and use cases. The authority remains open to issuing additional licenses in the future.

In May 2025, both the United States and Hong Kong accelerated efforts to legislate stablecoin regulation. In July of the same year, the U.S. passed the GENIUS Act, while Hong Kong’s Stablecoin Ordinance came into effect on August 1, 2025. On November 28, 13 Chinese government agencies, including the People’s Bank of China, reiterated their crackdown on cryptocurrency trading and classified stablecoins as virtual currencies, effectively ruling out their trading within mainland China.

Red Hat lays off its China R&D team

According to The Register, U.S.-based open-source software giant Red Hat has recently dissolved its China R&D team, relocating most engineering roles to India. The layoffs are expected to affect between 300 and 500 employees.

The move came abruptly. Employees claiming to be Red Hat engineers in China reported on forums such as Hacker News that their VPN access was suddenly cut off and internal system permissions revoked, followed shortly by termination notices. A leaked internal memo confirmed the decision. In the memo, Red Hat CTO Chris Wright stated that the company is shifting its R&D focus toward an “Asia-Pacific hub,” with India as a key investment region, and that the relocation would not reduce the company’s global R&D headcount.

Red Hat has long provided technical support to the U.S. military and secured an $848 million software contract with the U.S. Department of Defense in 2024. Relocating R&D operations may help mitigate national security scrutiny from Washington. Previously, Microsoft faced Pentagon criticism for involving engineers based in China in Azure projects supporting the U.S. military, and ultimately ceased using China-based staff for such work in 2025. Meanwhile, Red Hat’s parent company IBM now employs more people in India than in the United States.

Despite the complete withdrawal of its R&D presence, Red Hat has not halted commercial operations in China. Given China’s push for domestic IT alternatives and the open-source nature of many Red Hat technologies (such as CentOS derivatives), local vendors can still legally access and build upon its code.

The Linux kernel project introduces rules for AI-generated code submissions

The Linux kernel project has officially adopted a policy allowing AI-assisted code contributions, bringing months of heated internal debate to a close. Under the new rules, contributors using AI tools must include an “Assisted-by” tag to disclose such assistance, rather than using the legally binding “Signed-off-by” tag. Any bugs, security issues, or license violations arising from AI-assisted code will be the sole responsibility of the submitting developer.

For the open-source community, the originality of AI-assisted code remains a particularly thorny issue. Since AI models are often trained on code with restrictive licenses, developers cannot easily prove the legal origin of their contributions. Previously, undisclosed AI-assisted submissions had already sparked widespread backlash. Late last year, NVIDIA engineer and kernel maintainer Sasha Levin submitted patches generated by an LLM without disclosure, leading to performance regressions and strong protests. Around the same time, the GZDoom open-source project fractured after a core developer concealed the use of AI-generated code. Projects such as Gentoo and NetBSD have since banned AI-generated contributions entirely to avoid copyright risks.

In addition, the proliferation of AI tools has led to a surge in low-quality code and issue submissions, with projects like cURL and Node.js facing ongoing waves of spam-like contributions.

In discussions, the Linux founder adopted a pragmatic stance, arguing that AI is fundamentally just a tool and that outright bans are both ineffective and unenforceable. Instead, he emphasized the importance of accountability for human developers—a position that ultimately shaped the final policy.

Betting on weather becomes a trending trading strategy

According to Bloomberg, weather prediction markets are experiencing explosive growth. On platforms such as Kalshi and Polymarket, participants ranging from weather enthusiasts to AI companies are increasingly placing bets on specific events like snowfall and temperature changes. In January alone, a single contract tied to a U.S. snowstorm saw trading volume exceed $6 million.

This emerging market is also becoming a testing ground for weather tech companies to refine their AI models. Some firms encourage employees to participate in betting to identify data noise in official meteorological stations and improve forecasting algorithms, while others have established investment funds to arbitrage using their proprietary models. Meanwhile, retail participants with little meteorological background have reported substantial profits by betting on temperature outcomes in cities like New York and London.

The scientific community and insurance industry are also exploring customized weather prediction markets. Institutions such as French reinsurer SCOR are sponsoring markets where experts can bet on macro trends like El Niño or hurricane frequency, providing valuable pricing signals for the insurance sector.

Analysts suggest that prediction markets, driven by direct financial incentives for accuracy, may outperform traditional government forecasts. However, concerns are growing as climate change intensifies and extreme weather events become more frequent. Critics warn that turning weather forecasting into a form of gambling could encourage zero-sum speculation, data manipulation, or even deliberate interference with meteorological monitoring systems.

Cyberspace Administration and Railway Authority summon third-party train ticket platforms for talks

On April 10, the Cyberspace Administration of China announced that, in accordance with the Cybersecurity Law and the Regulations on the Security Protection of Critical Information Infrastructure, it, together with the National Railway Administration, recently summoned seven third-party internet platforms involved in train ticket sales, including Ctrip, Tongcheng, Qunar, Fliggy, Meituan, Zhixing Train Tickets, and High-Speed Rail Manager. The authorities required these platforms to strictly comply with relevant cybersecurity laws and regulations, and not to use automated programs to conduct large-scale, high-frequency ticket-grabbing operations that interfere with the security verification mechanisms of the Railway 12306 platform, nor to disrupt or endanger its stable and secure operation.

The Cyberspace Administration stated that relevant departments will strengthen technical monitoring going forward. Any use of technical means to interfere with or undermine the security of the Railway 12306 platform will be dealt with strictly in accordance with laws and regulations such as the Cybersecurity Law and the Regulations on the Security Protection of Critical Information Infrastructure.

The Railway 12306 technical team had previously stated that third-party ticket-grabbing platforms, through high-frequency requests, consume large amounts of server resources and bandwidth—effectively resembling DDoS attacks—which can slow system response times, cause lag, or even lead to system crashes. Ahead of the 2026 Spring Festival travel rush, 12306 announced upgrades to its anti-bot system, incorporating multi-dimensional analysis including access frequency, user behavior, device characteristics, account credibility, and network IP. Suspicious requests are placed into a slow queue for processing, with the system capable of intercepting tens of millions of abnormal access attempts per day.

In December 2025, the Beijing Municipal Administration for Market Regulation organized an administrative meeting with 12 platforms including Ctrip, Qunar, Fliggy, Tongcheng, Meituan, and High-Speed Rail Manager, focusing on misleading promotions such as “speed-up packages,” “dual channels,” and “ticket monitoring,” as well as implications that paid services could grant priority access to tickets, and required rectifications.

News Worth a Quick Look

• On April 10, YouTube Premium in the U.S. announced another price increase. The individual plan rose from $13.99 to $15.99, the family plan from $22.99 to $26.99, and the Lite plan—which removes only some ads—from $7.99 to $8.99. YouTube Premium had previously raised prices in the U.S. and multiple international markets in 2023 and 2024. Last month, Netflix and Amazon Prime Video also increased their prices.
• Mark Gurman claims—
  • Recent supply chain rumors suggesting that the foldable iPhone is facing production bottlenecks and may be delayed until 2027 are inaccurate. Apple is reportedly not encountering major mass production issues, and the device remains on track to debut in September, with a market release expected shortly thereafter.
  • Apple’s smart glasses, internally codenamed N50, are undergoing intensive testing. The device will not feature a display, instead relying on an integrated array of cameras and microphones to support photo and video capture, audio playback, and AI voice interaction. It is made from high-end acetate materials, and the design team is currently testing at least four frame styles, including slim rectangular, classic wide rectangular, and oval shapes. The product is planned for release in 2027.
  • Former Apple AI chief John Giannandrea is set to officially leave the company after his stock vesting period ends on April 15, one year after stepping back from leading Apple Intelligence due to underwhelming performance and repeated delays in Siri upgrades. He is expected to move into advisory or board roles at startups after his departure.
• According to CoinDesk, rising global energy prices driven by geopolitical tensions in the Middle East have put Bitcoin miners under significant pressure. Data from analytics platform Checkonchain shows that by mid-March 2026, the average production cost of one Bitcoin had risen to $88,000. In comparison, the market price hovered around $69,200, implying a loss of about $19,000 (21%) per coin mined.
• According to The Washington Post, Anthropic recently held a closed-door meeting at its San Francisco headquarters, inviting around 15 Christian leaders. Over the two-day event, discussions focused on guiding the ethical and spiritual development of AI, covering topics such as how chatbots should respond to grief or self-harm tendencies, and even whether AI could be considered a “child of God.” Attendees noted that Anthropic’s team expressed significant concern over the growing unpredictability of AI systems. Researchers studying internal model mechanisms recently suggested that systems like Claude may already exhibit “functional emotions.” Some participants stated they were unwilling to rule out the possibility that humans might bear moral obligations toward the AI systems they create. Religious and academic attendees viewed the initiative as an attempt by Anthropic to move beyond Silicon Valley’s traditionally secular mindset and seek ethical guidance from external belief systems. The summit is reportedly just the beginning, with Anthropic planning further engagements with representatives from other philosophical and religious traditions.