With Congress’s frontier-AI vehicle still in draft and the executive branch issuing no new rules, the week belonged to the states, and to governors in particular. Rhode Island’s Dan McKee signed three AI laws on June 22, led by a therapy-chatbot ban that treats AI-delivered psychotherapy as practicing without a license, while Arizona’s Katie Hobbs vetoed all three of her legislature’s AI bills three days earlier. The live federal story stayed the Anthropic export-control standoff: on June 26, Commerce partially lifted its directive, allowing Mythos 5 back for a vetted set of U.S. critical-infrastructure partners while keeping Fable 5 (the consumer model) fully suspended. California’s surviving bills cleared crossover committees, and the EU’s August 2 transparency deadline drew sharper compliance warnings. The signal for AI companies: chatbot safety is now the most durable category of binding state law, and government model-suspension power is a settled risk.
06.30.2026 The Framework Podcast
Rhode Island signed three AI laws and Arizona vetoed three, as Commerce partly restored Anthropic's Mythos 5. Chatbot safety is now hard state law.
The Framework: AI Policy Intelligence and Analysis Podcast
The Framework combines the utility of a legislative tracker with the insight of an editorial column. We track individual pieces of AI legislation and official global policy, stripping away the hype to analyze the hard facts.
The Framework combines the utility of a legislative tracker with the insight of an editorial column. We track individual pieces of AI legislation and official global policy, stripping away the hype to analyze the hard facts.Listen on
Substack App
Spotify
RSS Feed
Appears in episode
Recent Episodes











