Anthropic Proposes AI Safety Bill With Strict Death Threshold | Newsradio 600 KOGO

Anthropic, a major AI company, has proposed a draft law that sets strict rules for stopping the release of AI models. According to TechCrunch, the law would let federal agencies stop an AI model only if it poses a "foreseeable, imminent risk" of killing 50 or more people or causing more than $1 billion in damage. Even in these cases, a federal judge would have to approve the decision.

The bill proposes that independent experts registered with the U.S. Commerce Department check the possible risks of AI models. AI companies could pay these experts, as long as their payment does not depend on the results of their reviews. TechBuzz points out that this rule would give judges, rather than expert regulators, the power to make these decisions. This could delay urgent action.

Anthropic's proposal comes as more people are examining AI safety. The company has supported this plan since June, according to a spokesperson who spoke to Truthout. Experts in the industry have raised concerns about whether the strict limits and the need for a judge's approval would be effective in preventing risks from AI.

The proposal fits Anthropic's plan to help shape AI rules before Congress creates its own laws. It shows the company's effort to balance safety checks with giving the industry room to operate. The bill is still only a proposal and has not become law.