A bill on blocking AI has begun to be promoted in the U.S.

Since the summer, the company Anthropic has been promoting a bill in the U.S. Congress on regulating artificial intelligence systems.

As 1news.az reports, this was announced by the newspaper Politico, citing sources.

The bill allows the government to block the release or use of AI models. To do so, expert groups would have to prove that a system could cause $1 billion in damage or create a risk of "the death or serious injury of more than 50 people." The threat must be directly linked to the creation or use of biological weapons, a model escaping its developer's control, or a cyberattack.

It is proposed that such risks be identified by independent experts registered with the U.S. Department of Commerce. Based on their assessment, the agency would be able to seek a freeze of the project through the courts.

The document also obliges developers to create internal security departments, publish reports on the operation of models, protect employees from retaliation for disclosing violations, and promptly notify the U.S. Department of Commerce of any serious failures. The provisions of the document largely echo California's AI regulation laws approved last year, Politico notes.