Anthropic wants courts to suspend dangerous AI models, OpenAI wants labs to shut them down
In late October, Anthropic and OpenAI sent letters to California Governor Gavin Newsom outlining competing plans for an AI kill switch. Anthropic proposes that courts be empowered to suspend a model after an independent review flags a near‑term risk of catastrophic harm. OpenAI instead wants a statutory duty for developers to shut down unsafe systems themselves, amending existing law SB 53.
Key points
- Anthropic seeks court‑ordered suspension of dangerous models after independent review.
- OpenAI wants a statutory duty for labs to shut down unsafe systems, amending SB 53.
- California Executive Order N‑9‑26 sets Nov 16 deadline for recommendations on audits, safety reports and a kill switch.
Both letters respond to Executive Order N‑9‑26, signed Sept 18, which directs the state to deliver recommendations by Nov 16 on independent auditors, safety reports, a kill switch and incident reporting. The proposals differ on who holds the final authority: courts versus a legal duty on labs. The debate has practical implications for businesses running AI agents, as the entity controlling infrastructure may become legally responsible for pulling the plug.
The story so far
2 episodes →- Anthropic wants courts to suspend dangerous AI models, OpenAI wants labs to shut them downthis story
AI Kill Switch: Anthropic and OpenAI Split on Who Pulls the Plug
shiftdelete.net · 1 November 2026
Loading the full article…
This text was published by shiftdelete.net and written by Hakkı ALKAN. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Policy & Regulation
All →- OpenAI plans policy push to align with incoming Democratic Congress · 2 src
- AI stealing Indigenous Australian art, artist says · 1 src
- Anthropic restricts Claude Code from recreating specific video games · 2 src
- OpenAI safety team flagged teen planning school shooting, lawsuits allege inadequate response · 1 src
- OpenAI, Google, Microsoft, and Anthropic back shared robots.txt opt-out for AI training · 11 src
Comments
via GitHub Discussions