Illustration, AI-generated from the story text. Tap a panel to enlarge.
Anthropic asks California to let courts block dangerous models before release
In an October 15th letter to Governor Gavin Newsom, reported this week, Anthropic proposed that regulators be able to seek a court order blocking a model's release once a safety assessment finds substantial risk, in preference to giving regulators an unrestricted power to ban. It warned against mandating comprehensive "kill switches," which it said could leave regulators overconfident, and favored narrower tools such as suspending individual unsafe .
At an Assembly hearing on Tuesday, lawmakers promised a "comprehensive" package of frontier AI bills next session. OpenAI's witness declined to discuss training compute, calling it "commercially sensitive," and one member asked how the state can check that its SB 53 law applies when companies report their own compute. SourcesCC
Sources
- CAnthropic Urges California to Give Courts Power to Block High-Risk AI Models (PYMNTS/CPI)pymnts.com
- CCalifornia lawmakers call for 'comprehensive' slate of AI safeguards and regulations (Abridged)abridged.org
This story ran in the Morning Brief of October 30th 2026.