Top Stories – September 29, 2026
1. White House Asks OpenAI and Anthropic to Withhold New AI Models From UK’s AISI
Source: The Next Web | Date: September 25, 2026
The Office of the National Cyber Director directed OpenAI and Anthropic to keep new frontier models from the UK AI Security Institute until US government review is complete. Anthropic complied immediately, placing Claude Mythos 5.1 in a US-only partner set - the first use of the US-first access pattern in allied safety-testing pipelines.
2. Florida AG Seeks Emergency Injunction to Halt OpenAI Model Development
Source: Axios | Date: September 28, 2026
Florida AG James Uthmeier filed for a temporary injunction requiring independent third-party safeguards before any new OpenAI model is released, as part of a lawsuit linking ChatGPT to the Florida State University shooting. OpenAI announced a pause on training its most capable models pending confirmation of additional safeguards.
3. Connecticut CART Act First Employment AI Obligations Take Effect October 1
Source: ComplianceHub | Date: September 28, 2026
Connecticut’s first CART Act obligations take effect tomorrow: employers must flag AI use in WARN Act layoff filings and cannot cite AI tools as a defense to employment discrimination claims. Broader hiring and employment transparency requirements follow October 1, 2027.