A bipartisan group of senators introduced a bill Thursday that would create an AI Safety Review Office at the Commerce Department, a move aimed at supporting cutting-edge AI model developers.
The Sustaining American Preeminence in AI Act of 2024 (S.5616), sponsored by Sen. Mitt Romney (R-Utah), would require the agency, led by a Senate-confirmed Under Secretary of Commerce, to prepare and issue opinions. . Home testing for model developers to assess the “extreme risk” of a system before development. The bill also requires the new office to work with the National Institute of Standards and Technology’s AI Safety Lab to provide best practices and technical assistance to industry, according to the release. That’s what it means.
The bill is co-sponsored by Sen. Jerry Moran (R-KS), Sen. Angus King (R-Maine), Sen. Jack Reed, and Sen. Maggie Hassan (D.N.H.). — Also establishes reporting requirements and guidelines for frontier developers to prioritize the national security implications of AI. This approach is part of an effort to balance the need to compete with foreign adversaries, the release states.
Romney said in a release that he and his colleagues on the Senate Commerce, Science, and Transportation Committee have “spent the past several months working to ensure that cutting-edge models do not pose a risk while maintaining America’s competitive edge in AI.” “We have transformed the framework for this into bipartisan legislation.” Unchecked chemical, biological, radiological, nuclear, or cyber risks. ”
Hassan said: “We must continue to work to protect against potential risks to public safety, including the misuse of this technology by terrorists and criminal organizations, by implementing balanced oversight of cutting-edge AI models. , allowing us to maintain technological leadership while protecting public safety and the national interest.”
Additionally, the agency will study future AI-related risks and submit a report on its findings to Congress.
Meanwhile, frontier AI model developers will receive assistance from the Commerce Department to identify red teaming best practices for chemical, biological, radiological, nuclear, or cyber risks.
Under the law, developers can ask the AI
If the law were enacted, developers who do not comply with the law would face criminal and civil penalties. If criminally charged, he could be sentenced to up to 10 years in prison. The civil charges would include fines of up to $1 million per day.
The bill also requires infrastructure-as-a-service providers to implement “know-your-customer” standards for transactions with foreign nationals. And state-of-the-art data centers training cutting-edge AI models must report their ownership and facility location to the Commerce AI Office.
