Policy · Updated 21 Oct, 04:31 am IST
Anthropic proposes draft AI safety bill to let government block dangerous models
Why it matters for readers: The proposal explains one concrete way regulators might decide when an AI model is too dangerous to release.
- The draft bill would permit blocking a model if it could foreseeably cause the death of more than 50 people or over $1 billion in property damage.1
- Catastrophic risks must stem from creating or releasing a biological weapon, the model evading developer controls, or a cyberattack.1
- Anthropic proposes using 'independent evaluators' registered with the U.S. Commerce Department to analyze models' potential for disaster.1
- Evaluators could be paid by AI labs, provided payments are not contingent on the evaluation results.1
- The numeric thresholds mirror California's recently passed Senate Bill 53, which also defines similar 'catastrophic risk' criteria.1
Get a brief like this every morning
Uzha reads hundreds of sources and gives you the stories that matter for your work, with every source linked. Free.
Get started