Anthropic AI Bill Halts Models Only at 50 Deaths or $1B — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Anthropic is pushing a federal AI safety bill that would authorize the government to halt an AI model's release only if it poses a risk of $1 billion in damages or killing or injuring more than 50 people.
- The bill designates an independent third party registered with the Department of Commerce to periodically evaluate AI labs and models, with the lab permitted to pay the evaluator as long as independence is not compromised.
- The proposal makes the Commerce Department the sole federal authority empowered to order a model release halted, requiring a federal judge in the District Court for the District of Columbia to make that call within a 72-hour reporting window.
- Triggers for intervention include the potential creation of a bioweapon, AI-powered cyberattacks, or an AI lab losing control of a model or agents.
- Senator Bernie Sanders (I-Vt.) has introduced a competing bill proposing a 20-year prison term for anyone developing AI superintelligence, treating it on par with illegal nuclear weapon development.
- Congress is separately working on a proposal to enforce kill switches on the most powerful AI models, prompted in part by Anthropic CEO Dario Amodei's warning that AI-driven botnets could take over the internet within 6 to 12 months.
Why it matters: The bill's triggers — $1 billion in damages or 50 deaths — are set so high that intervention would only kick in after catastrophic harm is already nearly certain, giving labs a wide operational runway. Concentrating enforcement authority solely in the Commerce Department also strips every other federal agency of the power to halt a dangerous release, narrowing the regulatory safety net.
Ask SkimNews
