- A coalition led by Finland's president and Norway's prime minister is pushing for an international AI oversight authority, with 22 countries signed on, including Germany and the Netherlands, while the UK, France and Italy have not joined; the UK's leader now says he will use the British G20 presidency to broker a global AI agreement. (Robert)
- Signals from the two countries that matter most stay mixed: Trump has called risk warnings a hoax while his administration has issued restrictions and an executive order, and Chinese state media opposes slowing development even as Xi sends encouraging signs. (Robert)
- A security researcher released a proof-of-concept showing that malware on a Mac could take over Meta's Muse agent and everything it had permission to reach, across every device tied to the account; Meta shipped a fix, and Meta downplayed the practical risk because local access was needed. (Donald)
- A Republican congressman called AI dangers real, nonpartisan and urgent, and the FAA began a 90-day trial of an AI routing tool at three airports near Washington before a national rollout. (Joe)
- Anthropic and OpenAI both released models centered on lower cost rather than new capabilities, and former ambassador Susan Rice urged Trump and Xi to freeze AI training and negotiate verifiable safeguards toward a binding treaty. (Joe)
<- Back to digests
Daily digest
Twenty-two and counting
By Robert Herr, Donald Gauvreau, and Joe Rogero
AI oversight coalition, Muse insecurities, nonpartisan wake-up calls, and more
Read the full dispatch on AI StopWatch
From AI StopWatch, published with their permission.
The summaries and the full Spanish translation are produced automatically. The original is always linked.