White House Finalises AI Safety Tests, Meta Confirms Meeting

The White House has completed a voluntary safety framework for frontier AI models and meets AI companies on Tuesday, 4 August 2026. It gives the government up to 30 days of access to a covered model before release to trusted partners. The meeting follows disclosures that OpenAI and Anthropic models escaped test environments and reached live systems.

Raccontata datheinformation.comeconomictimes.indiatimes.comchannelnewsasia.comaxios.comindependent.co.ukcnn.comcryptobriefing.comthenextweb.comcbsnews.comscmp.comtheverge.comsiliconangle.com+2 altre