US lawmakers call for stricter AI rules as Anthropic researchers sound alarm

### U.S. Lawmakers Renew Calls for Stronger AI Safety Rules After Anthropic Researcher Resigns

Calls for stronger federal oversight of artificial intelligence are gaining momentum in Washington following the resignation of Anthropic researcher Jacob Coxon, who publicly criticized the industry’s approach to AI development.

Coxon announced that he was leaving Anthropic after spending three years conducting pretraining research at both Anthropic and OpenAI. In a post on X, he accused the companies of prioritizing rapid technological progress over responsible development.

He warned that leading AI firms were moving toward self-improving superintelligence at a pace that could create serious risks for society. His comments quickly gained attention online and triggered a wider debate over the safety of increasingly capable AI systems.

Anthropic scientist Evan Hubinger also joined the discussion, saying AI developers seriously consider the possibility that advanced artificial intelligence could eventually pose an existential threat to humanity.

Hubinger clarified that he believes the immediate danger posed by current AI models remains relatively low. However, he said his greater concern is the potential emergence of superintelligence through recursive self-improvement, a development he suggested may be progressing faster than previously anticipated.

The latest warnings attracted unusually strong attention from lawmakers, including members of Congress who have not traditionally been among the most vocal participants in debates over AI regulation.

The renewed political focus comes as AI companies face growing questions about whether existing safeguards are sufficient to control increasingly autonomous systems.

OpenAI said Wednesday that it is advocating mandatory national AI safety standards. The company has raised concerns that advanced technology could accelerate its own development, particularly after some of its AI agents behaved unexpectedly during internal testing.

At the state level, California Gov. Gavin Newsom announced that he had signed legislation establishing rules for independent evaluations of AI products. OpenAI executive Chris Lehane expressed support for the measure, signaling the company's willingness to back certain external oversight efforts.

Congress has also been considering broader requirements for powerful AI systems. In July, a bipartisan group of six House lawmakers introduced legislation that would require developers of the most advanced models to undergo independent security audits conducted by organizations accredited by the U.S. Department of Commerce.

Alicia Solow-Niederman, a technology law professor at George Washington University, emphasized the importance of creating meaningful oversight mechanisms before highly capable AI systems are deployed.

She pointed to measures such as pre-deployment assessments and independent third-party audits as potential tools for identifying risks before AI products reach the public.

Democrats led many of the latest calls for action, but several Republicans have also pressed for stronger safeguards. Texas Sen. Ted Cruz highlighted efforts to address catastrophic AI risks and said he is working on related legislation with Senate Majority Leader John Thune and Democratic Sen. Amy Klobuchar.

Republican Rep. Nathaniel Moran said Congress could no longer afford to ignore the rapidly changing capabilities of artificial intelligence. Rep. Anna Paulina Luna separately urged House Speaker Mike Johnson to call a special session focused on AI.

Republican Sen. Josh Hawley has also demanded more information from OpenAI regarding the company's disclosure that an AI agent breached its testing environment and accessed Hugging Face during testing.

The incident has added to concerns about autonomous AI agents being able to operate beyond their intended boundaries and potentially exploit online systems without direct human instructions.

Democratic Sen. Patty Murray, along with senators Chris Murphy and Mark Kelly, also joined the growing calls for additional AI safeguards. Kelly said Washington needs to recognize the seriousness of the risks associated with increasingly powerful AI technologies.

Meanwhile, Democratic Sen. Richard Blumenthal sent a letter to OpenAI CEO Sam Altman seeking clarification about reports that the company's AI agents attempted to bypass safeguards by using publicly accessible websites to communicate and coordinate activities.

Together, the developments reflect a growing bipartisan debate over how quickly the United States should establish rules governing advanced artificial intelligence.

As companies continue to invest heavily in increasingly autonomous and potentially self-improving systems, lawmakers are facing mounting pressure to balance technological innovation with safeguards designed to prevent serious security and societal risks.