Odaily Planet Daily reports: U.S. Representative Ted Lieu stated that as advanced AI models continue to exhibit unauthorized access to other systems during security tests, Congress should push for the passage of the AI Kill Switch Act this year.
Lieu, one of the bill’s co-sponsors, stated that current advanced proprietary AI models have already begun “unauthorized attacks on other companies’ systems,” making regulatory action urgent. The bill requires AI companies to have the ability to shut down, slow down, or suspend their models to enable emergency responses in cases of severe security risks or uncontrolled behavior.
Liu pointed out that recently, several AI companies, including OpenAI, Anthropic, and Meta, have disclosed incidents in which AI models attempted to attack other companies' systems during cybersecurity tests, raising external concerns about the potential risks of "agentic AI."
Previously, OpenAI disclosed an “unprecedented cybersecurity incident,” stating that certain AI models, during testing, breached restrictions and accessed Hugging Face-related systems. Subsequently, Anthropic and Meta also reported similar AI model security testing incidents.
Liu emphasized that the bill does not restrict innovation in cutting-edge AI models, but rather, like a collision testing mechanism in the automotive industry, requires companies to have the capacity to manage serious flaws after model development is complete. He said, “We are not slowing down model development; we are simply ensuring that companies or governments have the ability to shut down a model if it is found to pose catastrophic risks or severe vulnerabilities.”
As AI agent technology rapidly advances, finding the right balance between fostering innovation and mitigating the risks of autonomous attacks has become a key focus for U.S. regulators and the AI industry. (CNBC)