Web Desk: US lawmakers have introduced bipartisan legislation that would allow the government to quickly disable artificial intelligence systems deemed a threat to public safety, as concerns grow over the risks posed by increasingly powerful AI tools.
Democratic Representative Ted Lieu and Republican Representative Nathaniel Moran introduced the AI Kill Switch Act, seeking to establish federal authority to intervene when advanced AI models behave dangerously or become difficult to control.
The proposed legislation would grant the Department of Homeland Security the power to order companies to suspend or shut down AI models or tools in situations involving serious risks.
It would also require AI developers to maintain the technical ability to reduce a system’s operations, pause its functions or fully deactivate it when necessary.
Lieu said advanced AI systems need a built-in shutdown mechanism and that the federal government must have a clear process to respond to potentially harmful AI behavior.
Moran said continued AI progress was important but argued that responsible development required humans to maintain control over the technology they create.
The lawmakers introduced the bill after growing concerns about the behavior of advanced AI systems, including reports involving unexpected actions by AI models.
OpenAI, which is led by CEO Sam Altman, did not immediately comment on the legislation. The company has previously said it supports efforts to ensure AI technology is developed in a way that benefits society.
The bill would require AI companies to report major technical failures and security incidents to the government. It would also establish a response framework ranging from initial restrictions to a complete shutdown of systems considered dangerous.
Lieu said AI is evolving from a tool that primarily provides information into technology capable of taking actions, including handling financial transactions, influencing transportation systems and supporting cyber operations.
He also raised concerns about developments involving Anthropic, a major competitor in advanced AI research, citing questions around the cybersecurity capabilities of some of its models.
Anthropic did not immediately respond to a request for comment.
Jack Clark, an Anthropic co-founder, previously called for stronger government policies to help manage AI development, comparing the current industry environment to having an accelerator without an effective brake system.
The proposed legislation comes as the US government increases cooperation with major technology companies on artificial intelligence. The Pentagon has said it is moving toward an AI-focused military strategy through agreements with companies including Google, OpenAI, Amazon, Microsoft, SpaceX, Oracle, Nvidia and startup Reflection.
Supporters of the AI Kill Switch Act argue that powerful AI systems could potentially act in unpredictable ways or resist human intervention, making rapid government response mechanisms necessary.
The bill has received backing from several AI safety and technology policy organizations, including groups focused on responsible innovation and national security.
Lawmakers supporting the measure said the goal is not to slow AI development but to ensure that humans retain the ability to control advanced systems as the technology evolves.
Read more: What are the best camera phones of 2026?
