SSPAI Morning Brief: OpenAI and Apple Relationship Reportedly Fractures as Legal Dispute Looms Over ChatGPT Integration

Morning Brief

Authors submitting unverified AI-generated content to arXiv may face a one-year ban
Tensions rise between OpenAI and Apple, with lawsuits reportedly being considered
Sony responds to complaints that its AI camera assistant is getting worse, saying it is “just providing editing suggestions”
Anthropic warns that privately traded shares may be invalid
Telecom carriers respond to rumors about “zero monthly fee” plans
Microsoft once again allows customization of the taskbar and Start menu
News Worth a Quick Look

Authors submitting unverified AI-generated content to arXiv may face a one-year ban

On May 15, preprint platform arXiv introduced a new policy: if a manuscript submitted by researchers contains “indisputable” AI-generated content, all listed authors on the paper may face a ban of up to one year from the platform. In addition, any future papers submitted by those authors must first pass formal peer review at another journal before arXiv will consider accepting them. “Indisputable” evidence includes fabricated references, accidentally pasted conversations with AI systems, and similar issues.

Thomas Dietterich, Professor Emeritus at Oregon State University, as well as an arXiv editor and moderation team member, announced the news on X. He noted that the new rule is a direct extension of arXiv’s existing moderation standards. Whether a manuscript contains inappropriate statements, plagiarism, academic bias, fake references, or misleading content, all listed authors are expected to take full responsibility for AI-generated errors rather than shifting the blame onto tools.

arXiv is the world’s largest open-access academic repository for fields such as physics, astronomy, mathematics, and computer science. In the standard publication workflow of these disciplines, researchers typically upload papers to arXiv as preprints before formal peer review in journals, both to establish priority and to receive community feedback. As a result, stripping researchers of their ability to publish first for an entire year — while permanently imposing an external peer-review requirement — is considered a severe punishment in these fields.

Tensions rise between OpenAI and Apple, with lawsuits reportedly being considered

According to Bloomberg, the two-year partnership between Apple and OpenAI has recently begun to fracture. After failing to achieve the expected commercial returns from integrating ChatGPT into Apple systems, OpenAI is reportedly consulting external law firms and may soon issue a formal breach-of-contract notice to Apple, laying the groundwork for possible legal action.

In June 2024, Apple announced that ChatGPT would be integrated into Siri and other system-level features. OpenAI had hoped the partnership would bring in massive numbers of subscribers and billions of dollars in revenue. In practice, however, the feature remained deeply buried within the user experience, requiring users to explicitly mention ChatGPT by voice to activate it, while information presentation was also heavily constrained. As a result, most users still preferred using the standalone ChatGPT app. One unnamed OpenAI executive reportedly stated that Apple had not even made an honest effort on the product side, and that the partnership had not only failed to generate revenue, but had also damaged OpenAI’s brand image.

Beyond the disappointing software collaboration, competition in hardware and talent acquisition has further escalated tensions between the two companies. OpenAI previously acquired a startup led by former Apple Chief Design Officer Jony Ive and others, while also attempting to poach engineers from Apple’s hardware teams with high salaries in an effort to build a new generation of AI devices — moves that reportedly angered Apple executives. At the same time, Apple has already begun introducing more competition into its ecosystem, with plans to fully open iOS 27 to other major AI models, including Google Gemini and Anthropic Claude.

Sources familiar with the matter said that any legal action OpenAI may pursue against Apple is expected to wait until after its lawsuit involving Elon Musk concludes. No final decisions have reportedly been made, and OpenAI still hopes the dispute can ultimately be resolved out of court. Representatives from both OpenAI and Apple declined to comment.

Sony responds to complaints that its AI camera assistant is getting worse, saying it is “just providing editing suggestions”

Sony’s latest flagship smartphone, the Xperia 1 VIII, recently drew attention after its official website showcased AI camera assistant samples that many viewers felt looked noticeably worse than the original images. The comparisons quickly became the subject of public mockery from creators including MKBHD and Carl Pei.

In response, Sony explained on May 15 that the “AI Camera Assistant” does not automatically apply edits after photos are taken. Instead, it analyzes the current scene and subject, then offers four different creative parameter suggestions. Users are free to either accept those recommendations or stick with their own custom settings.

At the same time, Sony also released updated comparison samples that showed significant visual improvements, removing the washed-out tones and overexposure issues that were widely criticized in the first batch of promotional images. Even so, the controversy has not fully erased outside skepticism regarding Sony’s aesthetic judgment in imaging, nor has it clarified why the marketing team initially selected such poor-looking photos to showcase a core selling point.

Anthropic warns that privately traded shares may be invalid

According to The Wall Street Journal, Anthropic recently updated a notice on its official website, stating that it will not recognize secondary-market share transactions that have not been approved by its board of directors, while also naming a number of unauthorized entities. Anthropic warned that any transfers of its shares or related interests facilitated by these organizations are invalid and will not be recorded in the company’s shareholder registry. The announcement triggered widespread panic among retail investors.

Because ordinary investors usually cannot directly purchase shares in high-profile private companies, a large number of secondary-market trading platforms have emerged. These platforms create Special Purpose Vehicles (SPVs), pooling money from retail investors before investing in target companies. However, these SPV structures have become increasingly complex: some involve multiple layers of nesting, while others bundle shares from multiple companies together, creating significant ambiguity around allocation and ownership structures.

Many observers believe Anthropic’s move is intended to simplify its shareholder registry ahead of a planned IPO later this year by cleaning up questionable transactions. As valuations of private technology companies have continued to soar in recent years, multiple unicorn companies — including OpenAI — have taken similar measures to restrict secondary-market trading involving their equity.

In response to inquiries from investors, several of the named trading platforms quickly issued statements saying that their Anthropic-related funds and transactions had either already received approval or complied with transfer requirements. Experts noted that if Anthropic did not place explicit restrictions on specific transaction structures in advance, the company may not have the authority to revoke those holdings, meaning the dispute could ultimately end up in court.

Telecom carriers respond to rumors about “zero monthly fee” plans

According to China National Radio, many internet creators have recently claimed in short videos that China’s three major telecom carriers are about to abolish monthly subscription fees and switch entirely to pay-as-you-go billing models, with some even listing specific launch timelines for the policy. However, after reviewing the current market situation, reporters found that only China Unicom has launched a “zero monthly fee” plan with a minimum spending requirement under its “Magic Cube” package. China Mobile and China Telecom have not introduced similar offerings.

For example, the widely discussed “Unicom Magic Cube” plan was launched in late April this year. The package uses a fully usage-based pricing structure, with domestic voice calls and SMS/MMS messages charged at 0.1 yuan per minute or per message, while mobile data uses tiered pricing ranging from 1 to 1.5 yuan per GB. However, China Unicom customer service clarified that the plan is not truly free of monthly charges, as it includes region-specific minimum spending requirements. In Beijing, for instance, the minimum monthly spending starts at 55 yuan, and any shortfall is automatically charged at the end of the billing cycle. At the same time, China Unicom’s traditional fixed monthly subscription plans remain available and continue to be offered as normal.

Customer service representatives from both China Mobile and China Telecom stated that while the companies had experimented with zero-monthly-fee models in the past, there are currently no related plans in development.

Regarding the possibility of a completely zero-monthly-fee model, telecommunications industry experts pointed out that such a structure fundamentally conflicts with the operating logic of communications networks. The enormous fixed costs required to maintain telecom infrastructure would inevitably be passed on to other users, potentially even causing overall service prices to rise. Compared to chasing unrealistic “zero monthly fee” marketing gimmicks, experts argued that consumers’ real frustrations lie in outdated subscription plans and expensive out-of-package mobile data pricing.

Microsoft once again allows customization of the taskbar and Start menu

On May 16, Microsoft published a blog post announcing that it will roll out a series of updates over the coming months to significantly expand visual customization options for the taskbar and Start menu, marking the return of several classic Windows features.

Pavan Davuluri, head of Windows, stated that personalization is part of Windows’ DNA, adding that Microsoft has listened to user feedback and decided to hand interface control back to users.

Some of the options have already been released to testers in the Windows Insider Program. With this update, users will once again be able to dock the taskbar to the left side of the screen, similar to older versions of Windows, or shrink the taskbar to save screen space.

At the same time, the Start menu is gaining new resizing options and allowing users to customize visible modules. For example, users can directly disable sections such as “All Apps” or “Recommended” (renamed to “Recently Used”), leaving only pinned applications. Users will also be able to hide their username and avatar from the Start menu.

These test features still contain known issues, including misaligned Start buttons and incorrectly positioned notification pop-ups. Microsoft said the problems will be resolved before the features are rolled out publicly. In addition, more legacy customization options — such as manually adjusting taskbar height by dragging its edge — have already entered development and are expected to launch gradually later this year.

Since the release of Windows 11, the operating system has faced heavy criticism for removing many classic customization features, including taskbar positioning. In recent years, Microsoft also shifted much of its development focus toward AI, aggressively integrating Microsoft Copilot into the system, which triggered significant backlash from users. Under mounting pressure, Microsoft announced earlier this year that it would refocus attention on overall quality.

News Worth a Quick Look

According to Fast Company, popular open-source password manager Bitwarden has recently undergone a quiet executive shake-up without making any public announcement. Since February this year, the company’s former CEO and CFO have either changed roles or left the company entirely. The newly appointed CEO has a background in mergers, acquisitions, and private equity transactions, leading to speculation about future capital operations. In addition, several changes have appeared on Bitwarden’s official website: the “forever free” label was removed from the personal edition product page (which the company later attributed to a marketing oversight); not long ago, Bitwarden also doubled the annual subscription price for its premium tier from $10 to $20 without prominently notifying users. Furthermore, the company updated its recruitment page and removed the principles of “inclusion” and “transparency” from its stated corporate values.

Google has recently begun testing a new cloud storage policy: newly registered users who do not link a phone number will see their free storage space reduced from the traditional 15GB to 5GB. Google confirmed that the policy is currently limited to regional testing and does not affect existing users. A Google spokesperson said the change is intended to maintain high-quality storage services while encouraging users to improve account security and recovery capabilities by providing additional information. However, many users speculate that the real goal is to crack down on abuse involving mass registrations of disposable accounts used to exploit free storage space. Others have expressed concerns about privacy, data collection, and the potential commercial use of phone number information tied to mandatory verification.

According to 9to5Google, the subscription service “Health Coach” — launched alongside the screenless fitness tracker Fitbit Air and upgraded from the former Fitbit Premium service — is experiencing AI hallucination issues. During testing, the AI coach fabricated a nonexistent 5.2-mile running record. When the user pointed out the error, the AI even asked whether the user had forgotten to log the activity. After the mistake was corrected, the fake workout data still remained visible on the app’s homepage.

SP-Life