Rep. Ro Khanna (D-Calif.) is reportedly preparing to introduce an AI bill that would impose strict liability rules and bar “recursive” self-improving AI until the government establishes safety standards.
New Bill to Combat Potential Threat of Advanced AI
Khanna said in an interview with CNBC on Monday that there was a risk of “civilizational extinction” as well as “a safety risk of loss of control, and then there’s a misuse risk,” and both are required to be taken seriously.
A summary of the bill, which is known as the "Human Control Over AI Act," was seen by the publication. Under the bill, AI models would be prohibited from upgrading themselves or changing their own safety and shutdown settings. This restriction would remain in place until federal rules are set and a government agency approves the technology, it said.
In addition, the bill intends to form a new federal agency that would specifically watch over the advanced AI models made by companies like OpenAI, Anthropic, Alphabet Inc‘s (NASDAQ:GOOG) (NASDAQ:GOOG) Google DeepMind, and Space Exploration Technologies Corp‘s (NASDAQ:SPCX) SpaceXAI.
This new organization would be responsible for setting up safety rules, a licensing process for training and launching models, auditing of frontier AI models, and strict security rules to ensure human oversight and ongoing testing.
Third-party reviewers would also be stationed directly within every advanced AI lab to provide immediate reports to the agency. The organization would also mandate guidelines for isolated offline testing zones called sandboxes, emergency cutoff mechanisms like kill switches, and safeguards to keep models contained within the lab so they cannot spread online. In addition, the agency would oversee the AI chips used by the top-tier developers, implementing ways to regulate and monitor their application.
AI Safety Debate Heats Up
Earlier this month, Evan Hubinger, a researcher at Anthropic, put the odds of AI wiping out humanity within a decade above 10%, amid a broader industry safety debate. Last week, Anthropic CEO Dario Amodei and OpenAI’s CEO Sam Altman warned of the potential risks posed by AI at the United Nations Security Council, calling for greater international cooperation to manage increasingly capable AI systems.
Meanwhile, on Sunday, "The Big Short" investor Michael Burry took to social media platform X to warn about the consequences of increasingly recursive AI-generated content. He argued that the growing use of AI-generated training data could amplify errors while exposing a deeper limitation of large language models on the path toward artificial general intelligence.
Last week, Sen. Bernie Sanders (I-Vt.) urged President Donald Trump and Chinese President Xi Jinping to negotiate an international agreement to pause advanced AI development and prohibit superintelligence.
See More: Top Growth Stocks
Image via Shutterstock
Login to comment