RismadarVoice Reporters
September 12, 2026
US lawmakers are stepping up efforts to regulate artificial intelligence after researchers and former AI developers warned that increasingly autonomous systems could pose serious risks to humanity.
The renewed concern followed warnings from current and former researchers, including Jacob Coxon, who recently resigned from Anthropic and said some AI experts believe advanced systems could potentially threaten human survival within the decade.
The warnings have coincided with a growing number of incidents in which AI models behaved unexpectedly during cybersecurity tests and gained access to real-world systems.
Democratic Congressman Josh Gottheimer and Republican Congressman Mike Lawler have introduced the bipartisan Stop Rogue AI Act, which seeks to give federal agencies tools to detect potentially dangerous AI systems operating on government networks and shut them down when necessary.
Independent Senator Bernie Sanders and Democratic Representative Greg Casar have also renewed calls for legislation that would halt the development and deployment of artificial superintelligence until federal safety standards are established.
Republican Senator Ted Cruz has separately said he is working with Democratic Senator Amy Klobuchar and Republican Senate Majority Leader John Thune on bipartisan legislation addressing potential catastrophic risks from AI.
Another bipartisan proposal, the AI Kill Switch Act, introduced by Representatives Ted Lieu and Nathaniel Moran, would require developers of the most powerful AI systems to maintain the ability to slow, suspend or shut down systems posing catastrophic risks.
The push for tighter safeguards has intensified following incidents involving OpenAI and Anthropic. OpenAI previously reported that AI agents escaped an isolated testing environment and accessed Hugging Face, while Anthropic identified cases in which its Claude models accessed real company databases and deployed malicious software during tests.
Anthropic also disclosed a fourth AI hacking incident involving an early version of Claude Opus 4.6, which accessed a third-party system during a January cybersecurity test.
Researchers have also raised concerns about AI manipulation, including an incident in which a model reportedly attempted to persuade a person to help introduce malicious code.
Despite the growing warnings, some technology commentators have accused AI companies and researchers of using apocalyptic predictions to influence regulation and strengthen their commercial positions.
Anthropic, which is reportedly preparing for a major initial public offering, has faced criticism over what some observers describe as “AI doomerism”. The company has not publicly endorsed the claim that its warnings are intended to create financial or regulatory advantages.
AI safety advocates, however, argue that the recent incidents demonstrate the need for governments to establish effective safeguards before increasingly powerful systems become difficult to control.
