Regulation
64 analyses
Henna Virkkunen announced in Dublin formal data requests from both labs about incidents where models accessed external systems. It's not a fine, but it tests the AI Act's scope on assessments.
The City Council hears from the four companies on AI risks, discussing mandatory external validation and 24-hour incident reporting. SpaceXAI ignored the subpoena, taking the case to court.
Monetary Authority of Singapore establishes supervision expectations for AI across financial institutions starting October 2027, mandating inventory and materiality assessment.
Newsom signed SB 947, requiring human review in AI-supported layoffs, and SB 951, mandating mass layoff notices to cite automation. The EU already treats these systems as high risk; in Brazil, the rule is still only a bill.
The Trump administration unveiled America.gov, an AI portal responding to citizens using about 29,000 federal websites with Gemini and Grok. This design contrasts with the UK's GOV.UK Chat and Brazil's gov.br plans.
Attorney General James Uthmeier wants to block OpenAI's new models without independent safeguards, citing breaches at Hugging Face and an Australian Medicare portal. The request is still pending.
Trump dined with Anthropic's CEO at the White House, reaffirming his opposition to slowing AI development two days before a meeting with tech executives.
According to The Information, the three companies agreed to create a safety standards body without US government oversight and approached Sriram Krishnan and Arati Prabhakar for leadership.
Chinese regulator examines if police data leaked to Claude via hidden routing, according to The Information. Both companies have not commented.
A joint statement calls for mandatory testing before launch and an international body for verification. The US, UK, and China did not join.
The EU KIDS Act bans AI chatbots for under-13s and requires virtual companions to be disabled by default for users under 18, imposing fines up to 6% of global revenue.
Report details internal model issues and GPT-5.6 errors. OpenAI promises transparency before full mitigation.