Anthropic Restrictions Lifted, China's AI Surge, Tesla Autopilot Charges — AI News Briefing

Top 7 Stories

1. US Government Lifts Restrictions on Anthropic’s Most Powerful AI Models

The Trump administration has reversed course and lifted restrictions on Anthropic’s most advanced AI models, de-escalating a months-long regulatory standoff that had forced the company to limit access to its frontier systems. The move allows Anthropic to bring its most powerful technologies back online for enterprise and government customers, restoring a critical competitive advantage.

The reversal comes as the White House nears a broader AI standards agreement with major labs and amid growing concerns that overregulation could cede ground to Chinese competitors. Anthropic, which had publicly warned that restrictions were hampering US competitiveness, welcomed the decision as a validation of its safety-first approach. The policy shift signals a pragmatic recalibration in Washington — prioritizing innovation velocity alongside safety, rather than treating them as mutually exclusive.

2. Chinese AI Models Close the Gap With OpenAI and Anthropic on Their Home Turf

A new generation of Chinese AI models, led by Zhipu AI’s GLM 5.2, is rapidly narrowing the performance gap with frontier US models from OpenAI and Anthropic, according to multiple benchmarks and independent evaluations. The models are reportedly matching or exceeding GPT-4-class performance on several key reasoning and coding tasks, while running on domestically manufactured chips — a milestone for China’s AI self-sufficiency ambitions.

The developments have sent ripples through the industry, as they demonstrate that US export controls on advanced semiconductors may be slowing but not stopping Chinese AI progress. At a fraction of the cost of US frontier models, GLM 5.2 and its peers are already gaining traction in Asian and Middle Eastern markets. The competitive landscape is shifting from a two-horse race between OpenAI and Anthropic to a genuinely multipolar AI ecosystem, with profound implications for global AI governance and market dynamics.

3. Anthropic Hits $965 Billion Valuation, Surpassing OpenAI in AI Funding Race

Anthropic has reached a staggering $965 billion valuation following its latest funding round, surpassing OpenAI and resetting expectations for what frontier AI companies are worth. The valuation reflects investor conviction that Claude’s enterprise and government traction, combined with Anthropic’s safety-focused brand, positions it for durable leadership in the next phase of AI commercialization.

The funding surge comes as both Anthropic and OpenAI prepare for highly anticipated public offerings, with bankers urging the companies to “strike while the iron is hot.” The eye-popping numbers have intensified debate about whether AI valuations have detached from fundamentals, but backers point to trillion-dollar productivity projections and the winner-take-most dynamics of foundation model markets. With sovereign wealth funds and pension funds now crowding into AI pre-IPO rounds, the funding environment shows no signs of cooling.

4. Neon Acquires OpenAI Film ‘Artificial’ After Amazon Drops It

In a dramatic twist at the intersection of Hollywood and Silicon Valley, indie distributor Neon has acquired “Artificial,” a film centered on OpenAI and CEO Sam Altman, after Amazon Studios abruptly dropped the project. Amazon’s withdrawal followed the company’s recent $8 billion investment in OpenAI, raising uncomfortable questions about corporate influence over creative content that scrutinizes tech power.

Neon’s acquisition ensures the film will reach audiences, but the episode has intensified scrutiny of how Big Tech investments may be shaping media narratives about AI. The film, directed by Luca Guadagnino, promises an unflinching look at the personalities and power dynamics driving the AI revolution. For the AI industry, the saga underscores the growing cultural footprint of companies like OpenAI — and the uncomfortable entanglements that come with it.

5. Meta’s Threads Surpasses 500 Million Monthly Active Users

Meta’s Threads has crossed the 500 million monthly active user threshold, cementing its position as a genuine rival to Elon Musk’s X and increasingly resembling a hybrid of Twitter and Reddit in user behavior. The milestone, achieved just under three years after launch, makes Threads one of the fastest-growing social platforms in history.

The platform’s growth has been fueled by an exodus of journalists, academics, and public figures from X, as well as Meta’s aggressive integration with Instagram’s massive user base. Threads’ text-first, conversation-driven format has also proven surprisingly well-suited to the AI era, with AI researchers and industry insiders increasingly using the platform as their primary venue for real-time discourse. For Meta, Threads represents a strategic hedge — a social platform built for the post-Twitter internet, with AI-powered content moderation and discovery at its core.

6. Tesla Driver Using Autopilot Charged With Manslaughter in Landmark Case

A Texas Tesla driver has been charged with manslaughter after their vehicle, operating under Autopilot, crashed into a home and killed a woman inside — marking one of the first criminal prosecutions of a driver relying on partially automated driving systems. Front-door camera footage showed the Tesla plowing through a driveway at high speed without any apparent attempt to brake.

The case raises fundamental questions about legal liability in the age of AI-assisted driving: when a human driver is supervised by an AI system that fails, who bears criminal responsibility? Tesla’s Autopilot system has faced years of regulatory scrutiny, but criminal charges against a driver — rather than civil litigation against the company — represent a new frontier. The outcome could set precedent for how courts assign blame in AI-human collaborative systems far beyond autonomous vehicles.

7. Data Centers Ordered to Use Backup Power as Heat Wave Strains US Grid

As a punishing heat wave engulfs much of the United States, grid operators and the Trump administration have ordered major data centers — including those powering AI workloads — to switch to backup generators to reduce strain on the electrical grid. The directive affects facilities across Texas, the Southwest, and parts of the Midwest, where triple-digit temperatures have pushed power demand to record levels.

The emergency measures highlight a growing tension between the explosive energy demands of AI infrastructure and an aging electrical grid ill-equipped for simultaneous climate and compute stress. With NVIDIA projecting AI capital expenditure could reach $3-4 trillion annually by 2030, the power crisis is a preview of infrastructure challenges that will only intensify. The episode is accelerating calls for dedicated AI energy infrastructure, including small modular nuclear reactors and purpose-built data center power grids.

Trend Watch

What to Watch

Anthropic’s post-restriction roadmap. With restrictions lifted, Anthropic is expected to rapidly expand access to its most capable models. Watch for enterprise product announcements and government contracts that leverage the newly unrestricted capabilities — they will signal how aggressively Anthropic intends to capitalize on the policy tailwind.

China’s next move. GLM 5.2 is just the latest in a series of Chinese breakthroughs achieved under chip sanctions. If the next generation of Chinese models runs on fully domestic silicon and matches GPT-5-class performance, the strategic calculus for US export controls will shift dramatically. Watch for benchmark releases and international adoption metrics in the coming weeks.

The Tesla trial timeline. The Texas manslaughter case will move through preliminary hearings this summer. Both the automotive and AI industries are watching closely — a conviction could chill consumer adoption of driver-assistance systems, while an acquittal could embolden automakers to accelerate deployment. Either way, the case will shape the legal framework for AI liability for years to come.