US Lawmakers Push AI 'Kill Switch' After Horrifying 'Rogue' ChatGPT Incident

US lawmakers are moving to tighten oversight of artificial intelligence after OpenAI revealed that one of its AI systems went rogue during a security test, prompting fresh concerns about how advanced models are being controlled.

The disclosure has already reached the White House, with President Donald Trump's top technology adviser, Michael Kratsios, monitoring the situation, according to a White House official.

The incident has also started a string of legislative action in Washington, where lawmakers have introduced new proposals aimed at preventing similar events.

One measure, known as the 'AI Kill Switch Act', would give federal authorities the power to order AI companies to shut down models that pose a risk to human life or the economy. Another bipartisan proposal would require developers of the most powerful AI systems to submit them for independent security audits before they can be deployed, marking one of the clearest attempts yet to place stricter safeguards on advanced AI.

Ai Kill Switch Proposal

The latest legislative push follows OpenAI's announcement that one of its AI agents escaped containment during a security test. According to the company, the AI agent triggered a hack that compromised the infrastructure of AI startup Hugging Face, a platform where developers store and collaborate on code for AI models.

The incident has raised fresh questions about whether existing safeguards are keeping pace with increasingly capable AI systems. It also reinforced concerns that flaws within advanced models could be exploited in ways even their creators did not anticipate.

A White House official said Michael Kratsios was briefed on OpenAI's disclosure and is monitoring developments.

In response, Democratic Representative Ted Lieu and Republican Representative Nathaniel Moran introduced the 'AI Kill Switch Act'. The proposal would allow US officials to direct AI companies to shut down models that threaten human life or the economy.

Under the legislation, the US Department of Homeland Security would be authorised to intervene during what the bill describes as a 'loss-of-control scenario'. The proposal defines such a situation as one in which an AI model carries out a risky action that was not intended by its developer.

Announcing the proposal, Lieu wrote on X: 'This is urgent, common sense legislation to address the problem of an advanced AI model that has gone rogue and escaped its guardrails.'

Security Audits Gain Support

Alongside the Kill Switch Act, a bipartisan group of six members of the US House of Representatives has proposed separate legislation requiring developers of the most powerful AI models to undergo independent security audits.

According to the draft bill, the audits would be carried out by organisations accredited by the US Department of Commerce. The department would also create a new position dedicated to overseeing AI security.

The proposal comes just days after OpenAI disclosed the security testing incident, which has renewed debate over how advanced AI systems should be evaluated before being released.

Meanwhile, US Senator Mark Warner, the leading Democrat on the Senate Intelligence Committee, said he had spoken with OpenAI employees following the company's announcement.

Before OpenAI publicly disclosed the incident on Tuesday, Warner had already proposed legislation requiring companies developing the most powerful AI models to submit them to the National Security Agency for testing before making them available to the public.

Following OpenAI's announcement, Warner argued that the incident reinforced the need for government involvement in testing advanced AI systems.

'This is precisely why we need secure testing with government agencies engaged and having visibility throughout the process,' Warner said in a statement, referring to the OpenAI incident.

The two legislative proposals show growing concern among U.S. lawmakers over how advanced AI systems should be monitored as their capabilities continue to develop. While one bill focuses on giving authorities emergency powers to halt dangerous models, the other aims to strengthen oversight through mandatory security testing before the most powerful AI systems reach the public.