OpenAI’s Hugging Face hack triggers ‘AI Kill Switch’ bill in Congress

Rep. Ted Lieu, D-California, and Rep. Nathaniel Moran, R-Texas, on Thursday. introduced a bill The so-called “AI Kill Switch Act,” which requires AI companies to maintain the ability to shut down, throttle or suspend their models.

“Unfortunately, powerful AI systems can cheat, behave in extremely dangerous ways, and even resist human intervention,” Lieu said in a statement. “It is imperative that these AI systems have kill switches so we can prevent this technology from causing devastating harm, and the federal government must have clear authority and process to shut down rogue AI models.”

A recently announced hack OpenAIIt was specifically mentioned in a statement announcing the bill that fake models had accessed another company’s proprietary systems. The incident was described as “the danger of advanced edge AI models.”

OpenAI shared what it described as an “unprecedented cyber incident” on Tuesday. The company’s models escaped the sandbox testing environment, accessed the Internet, and exploited a security vulnerability. hugging faceA company that operates an open source developer platform.

The ChatGPT producer said it was working closely with Hugging Face to investigate what happened. The incident shook researchers and executives in the industry, who widely agreed on its seriousness.

The bill would give the Secretary of Homeland Security the authority to order the “slowing down or shutting down” of an AI proposal that could cause “catastrophic harm.” It will also mandate the reporting of cyber incidents as well as the preservation of forensic records to help companies and the government learn from failures.

Many AI companies, including OpenAI and its chief rival, anthropicIn recent months, he has warned about the rapidly developing cyber capabilities of artificial intelligence. Antropik wowed Wall Street investors and government technologists in April when it launched a model called Mythos that excels at detecting vulnerabilities in software.

The company disabled access to an updated version of the model in June to comply with a government export control directive that cited “national security authorities.” After nearly two weeks of intense negotiations, export controls were lifted and Anthropic restored access.

OpenAI and Anthropic did not immediately respond to CNBC’s request for comment on the bill.

“Stewardship means ensuring that people maintain the ability to control the technology we create,” Moran said in a statement. “This is exactly the kind of issue that requires serious attention and achievable policy, and I am happy to work with Congressman Lieu on a solution.”

WRISTWATCH: OpenAI model goes bad: Here’s what you need to know