Anthropic proposes draft legislation on federal artificial intelligence safety regulations

Artificial intelligence startup Anthropic has circulated draft safety legislation to United States lawmakers, TechCrunch and Politico reported. TechCrunch, Futurism, Cybernews, Mezha, and Truthout all reported that the draft bill sets specific thresholds before federal authorities can step in to halt an artificial intelligence model. [1][2][3][4][5]

Under the proposed draft, federal agencies would be prevented from blocking the release of an artificial intelligence model without permission from a judge, according to TechCrunch. TechCrunch, Truthout, Futurism, Cybernews, and Mezha all report that the policy establishes a scenario where a model could be blocked if it risks killing more than 50 people. Mezha further reported that the proposal includes a financial threshold of 1 billion dollars in damages. [1][2][3][4][5]

Anthropic operates under a governance model that includes an independent Long-Term Benefit Trust empowered to elect directors and safeguard its public benefit purpose. In 2022, the company introduced Constitutional AI, a training technique using natural-language principles rather than relying exclusively on human preference data to train its Claude models. The proposal arrives amid widespread debate over artificial intelligence safety, alignment, ethics, and technological unemployment during the recent accelerated development boom. [7][8]

Separate financial reports from Futubull Community state that Anthropic generated 16.8 billion dollars in third quarter revenue, representing a quarterly surge of nearly 50 percent. Congress has not yet announced whether it will formally consider adopting the draft legislation. [1][6]