AI safety company Anthropic has been discreetly pitching a draft bill to U.S. lawmakers aimed at regulating the safety and security of frontier AI models, Politico reported on Friday, a story subsequently covered by Reuters.
According to the report, the proposed bill would grant the government authority to block the release or use of an AI model only if it poses a "foreseeable, imminent risk" of causing more than $1 billion in damages or killing or seriously injuring more than 50 people.
The bill specifies that such a high-stakes risk must stem from one of three scenarios: the creation or release of a biological weapon, the model evading its developer’s controls, or a major cyberattack.
An Anthropic spokesperson told Reuters that the draft bill reflects proposals from the company's "Advanced AI Framework," which was released earlier this year. The spokesperson said the goal is to give the government the "ability to block or deter dangerous deployments of the most powerful models" and to mandate that developers test their models, be transparent about the results, and have them independently evaluated.
Under the proposal, independent evaluators would register with the Commerce Department and be tasked with periodically assessing AI labs and their frontier models. If an evaluator determines an "imminent enumerated risk" exists, they must report it to the Commerce Department within 72 hours.
While the Commerce Department could then ask a federal judge to halt a model's deployment, the proposed legislation explicitly prohibits any other government agency from suspending its development or use. This authority would rest solely with a judge in the District Court for the District of Columbia.
Source: Reuters, citing Politico
🌐
