National Economic Council Director Kevin Hassett told Fox Business on Wednesday the administration is studying a possible executive order to require AI models to be "proven safe, just like an FDA drug" before release, The Hill reported. His comments are the first on-record White House framing of the prospective EO around an FDA-style framework. Hassett said the administration mounted a whole-of-government effort with private-sector partners to test Anthropic's Mythos vulnerability-finding model before broader release. The framing follows Tuesday's CAISI announcement of binding pre-release testing pacts with Google DeepMind, Microsoft and xAI, as reported by NIST in AIPD's May 6th edition.
Read at The Hill ↗
Anthropic announced Wednesday it will use the entire compute capacity of SpaceX's Colossus 1 data center in Memphis, Tennessee, gaining 300 megawatts of new capacity and access to more than 220,000 Nvidia processors within the month, the Wall Street Journal reported. Elon Musk said on X he agreed to lease the compute after meeting Anthropic leadership the prior week, reversing his February characterization of the company's AI as biased, Reuters reported. Anthropic also said it is interested in working with SpaceX to develop multiple gigawatts of orbital data centers. Anthropic remains designated a Pentagon supply chain risk and is suing the Defense Department over its exclusion from the eight cleared AI vendors named by AI Business in AIPD's May 5th edition.
Read at WSJ ↗ • Read at Reuters ↗
White House Chief of Staff Susie Wiles said in a Wednesday statement on her new X account that the U.S. government would refrain from choosing winners and losers in artificial intelligence as the administration prepares new AI policy directives, Bloomberg reported. The statement followed NEC Director Kevin Hassett's Wednesday Fox Business interview pitching an FDA-style pre-release review, separately reported by The Hill. The Verge's Tina Nguyen reported Wednesday that Wiles and Treasury Secretary Scott Bessent are leading the prospective AI EO inside the White House, displacing former AI and crypto czar David Sacks. Their April meeting with Anthropic CEO Dario Amodei overrode earlier Pentagon objections that had labeled the company "woke" and pushed for its ban from government use.
Read at Bloomberg ↗ • Read at The Verge ↗ • Read at X.com ↗
The European Parliament and the Council of the EU reached political agreement Thursday on the Digital Omnibus on AI, the European Commission announced. The package locks compliance dates for high-risk AI systems used in biometrics, critical infrastructure, education, employment, migration, asylum and border control to Dec. 2, 2027. Rules for AI integrated into products such as lifts and toys take effect Aug. 2, 2028. The agreement also bans "nudification" apps as part of the package's citizen protection track.
Read at European Commission ↗
The U.S. Cybersecurity and Infrastructure Security Agency, the National Security Agency and Five Eyes counterparts (UK NCSC, Australia ACSC, NZ NCSC, Canadian Centre for Cyber Security) issued joint guidance Thursday titled "Careful Adoption of Agentic AI Services," Reseller News reported. The document defines agentic AI risk categories: privilege escalation, insecure tool integrations, rogue agents, cascading system failures and accountability gaps. One scenario describes a compromised procurement agent using excessive privileges to modify contracts and approve payments without detection. The agencies advise organizations to limit agentic deployments to low-risk and nonsensitive tasks until standards mature, and to apply least-privilege access, defense-in-depth strategies and continuous monitoring.
Read at Reseller News ↗ • Read at CISA ↗
A government ethics group accused two AI investor-backed super PACs of improperly concealing which companies they pay to produce advertisements and voter outreach, Bloomberg reported. The Campaign Legal Center's FEC complaint targets American Mission and Think Big. The complaint alleges each has routed more than 90% of its spending through two corporations registered in Delaware, totaling over $10 million combined this year. The filing says the PACs violate FEC disclosure rules requiring identification of the vendors performing the underlying campaign work.
Read at Bloomberg ↗