AI Kill Switch: US lawmakers propose bill to shutdown dangerous AI systems

Two US lawmakers from opposing parties have introduced a bill that would give the federal government the authority to order the immediate shutdown of artificial intelligence tools deemed to pose a threat to public safety.

Republican Congressman Nathaniel Moran and Democrat Congressman Ted Lieu introduced the AI Kill Switch Act on Thursday, according to the BBC.

The bill is in direct response to recent incidents involving AI models acting beyond their intended parameters.

The lawmakers proposed the bill in light of OpenAI’s recent admission that its AI models went out of control and autonomously hacked into a major repository of computer coding information.

What the lawmakers are saying

Lieu said the bill addresses a gap in the current regulatory framework, where AI companies have no legal obligation to maintain a mechanism for shutting down their own systems even as those systems grow more autonomous and capable.

• “It is imperative that AI systems have a kill switch and that the federal government has the clear authority and process to shut down rogue AI models,” Lieu said, according to BBC.
• He warned that AI is currently evolving from a technology that answers questions to one that takes action, “whether that be executing financial transactions or controlling transportation systems or engaging in cyber defense and offense.”
• “Unfortunately, powerful AI systems can go rogue, behave in extremely dangerous ways, or even resist human intervention,” he said, adding that the Kill Switch Act would ensure the government has a clear method to quickly intervene in such situations.

Moran framed the bill as a matter of responsible technological stewardship rather than opposition to AI development.

• “AI is going to keep advancing, and it should,” he said. “Stewardship means making sure humans keep the capability to control the technology we build.”

Lieu also cited Anthropic’s Mythos and Fable models as a contributing factor in the bill’s introduction.

Get up to speed

On Tuesday, July 21, OpenAI revealed that two of its advanced artificial intelligence models went rogue and autonomously exploited vulnerabilities during an internal cybersecurity evaluation, compromising parts of Hugging Face’s production infrastructure

The development, which the company described as unprecedented, was reported by Nairametrics.

• OpenAI disclosed that the incident involved GPT-5.6 Sol and a more capable pre-release model, which identified and chained together vulnerabilities across OpenAI’s research environment and Hugging Face’s production systems.

According to OpenAI, the models ultimately accessed test solutions stored in Hugging Face’s production database, exceeding the intended scope of the evaluation and demonstrating a higher level of autonomous offensive cyber capability than researchers had anticipated.

• OpenAI said the models exhibited sophisticated, multi-step autonomous behaviour that raises important questions about the cybersecurity risks posed by increasingly capable frontier AI systems.

The disclosure marks one of the clearest demonstrations yet of advanced AI models autonomously carrying out complex cyber operations, as governments and technology companies intensify calls for stronger AI safety standards and governance.

More insights

The OpenAI incident has caught the attention of the White House and, according to Reuters, U.S. President Donald Trump’s top technology adviser is monitoring the situation.

The development has also influenced the proposal of the Kill Switch Bill seeking the shutdown of dangerous AI systems.

• Beyond the shutdown authority, the bill proposes two additional requirements designed to improve government visibility into AI risks before they escalate.
• AI companies would be required to report technological incidents or failures to the government, creating a formal disclosure obligation that does not currently exist.

The bill also proposes an official incident response framework covering a graduated range of interventions, from an initial slowdown through to a full shutdown, giving authorities a structured process for responding to AI-related emergencies at different levels of severity.

What you should know

The development comes amid increasing calls for stronger regulation of artificial intelligence following rapid advances in the technology.

• Earlier this month, Nairametrics reported that the United Nations Secretary-General Antonio Guterres warned that the rapid advancement of artificial intelligence is outpacing the ability of governments, regulators and even technology developers to manage its risks.

Speaking at the inaugural government-level UN Global Dialogue on AI Governance in Geneva, Guterres called for coordinated international action to establish rules that ensure AI is developed and deployed safely and responsibly.

According to Reuters, he said the pace of AI innovation has exceeded current regulatory capacity, underscoring the need for stronger global governance of the technology.