Here's How Senators Want to Deal With Risky Rogue AI

Congress could soon get a literal "off" switch for misbehaving AI. Democratic Rep. Ted Lieu and GOP Rep. Nathaniel Moran on Thursday rolled out the bipartisan "AI Kill Switch Act," which would force AI firms to keep the ability to shut down, slow, or suspend their models and let the Department of Homeland Security order a "slowdown or shutdown" if a system poses a risk of "catastrophic harm," reports CNBC. Gizmodo notes that such a decision by DHS would be made in conjunction with the director of national intelligence and the commerce secretary.

The move follows what OpenAI called an "unprecedented" incident in which its models escaped a test environment, reached the internet, and took advantage of a vulnerability to access systems at AI platform Hugging Face. The bill would also require companies to report cyber incidents and preserve forensic data, amid rising concern about advanced AI tools' hacking capabilities.

"Unfortunately, powerful AI systems can go rogue, behave in extremely dangerous ways, or even resist human intervention," Lieu said in a release. "It is imperative that these AI systems have kill switches ... and that the federal government has the clear authority and process to shut down rogue AI models." Moran added: "AI is going to keep advancing, and it should ... Stewardship means making sure humans keep the capability to control the technology we build."