The White House is negotiating with Capitol Hill to secure federal preemptionPreemptionWhen federal law overrides state law on the same subject. In AI policy it refers to proposals, in Congress or by executive action, to block or freeze the growing patchwork of state AI laws in favor of a single national standard. of some state AI laws in exchange for tech policies on kids safety and deepfakeDeepfakeSynthetic audio, images or video that realistically depict a real person saying or doing something they did not. Non-consensual intimate deepfakes and election deepfakes are the two areas where states and Congress have actually passed laws. protections, The Hill reported. Sen. Marsha Blackburn (R-Tenn.) is leading the talks, with the package including the Senate version of the Kids Online Safety Act, the NO FAKES Act and age verification requirements, per Axios. Her spokesperson said the proposal is not "blanket preemption of all laws regulating AI or kids safety." The maneuvering indicates the Obernolte-Trahan Great American AI Act is no longer the likely vehicle for AI policy this Congress. The administration is also meeting AI companies this week on benchmarking for the June 2 voluntary pre-deployment executive order.
Read at The Hill ↗ • Read at Axios ↗
The White House has ordered the Center for AI Standards and InnovationCAISIThe Center for AI Standards and Innovation, the US government's AI testing body inside NIST at the Commerce Department. Formerly the US AI Safety Institute, it evaluates frontier models with developers on a voluntary basis and works with the UK's AI Security Institute. (CAISI), the Commerce-based unit that tests frontier AI models, to stop releasing public assessments, the Wall Street Journal reported. The pause, communicated by officials including National Cyber Director Sean Cairncross, will hold while last week's AI executive order takes effect. Cairncross and Treasury Secretary Scott Bessent had lobbied to put national security at the center of model evaluation, and the halt reflects their growing influence. Companies including OpenAI have urged the administration to preserve CAISI's role.
Read at WSJ ↗
Anthropic on Tuesday released Claude Fable 5, the first publicly available version of its Mythos model, with hard safety limits that route queries about cybersecurity, biology, chemistry and distillationDistillationTraining a smaller or cheaper model on the outputs of a larger one so it inherits much of the larger model's ability. It is a standard technique, but it has become a trade dispute because it lets a lab reproduce a rival's capabilities without matching its compute spend. to Claude Opus 4.8, the Wall Street Journal and TechCrunch reported. Anthropic said an external bug bounty produced no universal jailbreaks across more than 1,000 hours of testing before release. Pro, Max, Team and seat-based Enterprise plans get Fable 5 at no extra cost through June 22, after which it will require usage credits. Anthropic plans to restore it as a standard subscription feature. With the launch Anthropic will require 30-day retention on all traffic, including for enterprises previously holding zero-retention agreements, to defend against complex jailbreaks.
Read at TechCrunch ↗ • Read at WSJ ↗ • Read at Anthropic ↗
Taiwan authorities are considering stricter export controls on AI chip sales to China to align with U.S. measures and address semiconductor smuggling, the Straits Times reported, citing people familiar with the matter. The proposed controls would restrict sales to all customers in China, not just specific companies on Taiwan's existing export blacklist, and would let Taipei prosecute AI chip smuggling as a criminal violation for the first time. The move is part of ongoing U.S.-Taiwan trade talks under President Lai Ching-te's administration, per Bloomberg. Taiwanese authorities made their first known detentions of alleged AI chip smugglers in May on charges of falsifying documents. Gigabyte shares fell as much as 3.8% and Asustek 6.2% in early Taipei trading after the news.
Read at Straits Times ↗ • Read at Bloomberg ↗
European Commission regulators on Tuesday ordered Meta to restore access to WhatsApp for rival AI chatbot makers under interim measures while the bloc's antitrust investigation continues, per AP. The Commission said the action was needed to prevent harm to competition in the AI assistant market before its investigation concluded. The order remains in effect until June 2029 or until the probe closes, and noncompliance could trigger fines of up to 10% of Meta's annual revenue. Meta said it will appeal, calling the action "regulatory overreach subsidized by the many European companies that pay."
Read at AP ↗ • Read at FT ↗