WASHINGTON—Mitt Romney (R-UT), Jack Reed (D-RI), Jerry Moran (R-KS), Angus King (I-ME), and Maggie Hassan (D-NH). A U.S. senator delivered the first speech of its kind today. This is a law that deals solely with protecting the national security of the United States from extreme risks such as chemical, biological, radiological, nuclear, and cyber threats posed by future developments in advanced AI models. The bipartisan Maintaining America’s Advantage in AI Act, which builds on the group’s previously announced framework, would create an Office of Artificial Intelligence Safety Review within the Commerce Department. The agency will work with industry to protect frontier models from abuse by foreign adversaries who seek to exploit this technology.
The agency’s mission would be to prioritize the national security implications of AI while ensuring the U.S. domestic AI industry maintains an advantage over foreign adversaries. This law is limited to frontier models, i.e., cutting-edge AI models that have not yet been developed.
“While AI has the potential to dramatically improve and transform the way we live, it also comes with significant risks to national security. We have to protect it from our adversaries who want to inflict it on us,” said Senator Romney. “My colleagues and I have spent the past several months ensuring that our most advanced models do not pose unchecked chemical, biological, radiological, nuclear, or cyber risks, while preserving America’s competitive edge in AI. We have worked to transform the framework into bipartisan legislation that will help Congress and the next administration address the extreme risks posed by AI without hindering American innovation. I hope this will be a starting point for discussion.”
“AI offers tremendous benefits, but we must recognize the risks and take steps to ensure this tool is not misused by adversaries to threaten U.S. security and leadership. This bipartisan bill would establish an AI Safety Review Office within the Department of Commerce to ensure the United States keeps pace with technological advances and responsibly supports innovation, opportunity, and discovery.” Senator Reed said.
“Advances in artificial intelligence present opportunities for American innovation, efficiency, and strategic advantage,” said Senator Moran. “But we must leverage the power of AI responsibly to ensure we reduce the extreme risks to national security. We must seek feedback from across the technology industry, government, and other stakeholders. Following this, my colleagues and I developed a plan to help certain innovators continue developing AI to ensure a competitive edge against adversaries in this critical time, while also mitigating national security risks. We have created a new bill for the technology sector.
“In an evolving global threat landscape, the United States must stay one step ahead of new technologies to protect both national security and domestic and international interests, and we must carefully, cautiously, and thoughtfully leverage artificial intelligence.” It means moving into the future,”’ Sen. King said. “The bipartisan Sustaining America’s Advantage in AI Act provides important guidelines for federal oversight of AI technology to prevent it from being misused by bad actors seeking to cause harm. We need to ask important questions now to wisely guide our next steps and decisions. Collaborate on solutions that advance America’s progress while protecting our people from biological, chemical, cyber, and nuclear threats. Thank you to my colleagues for this.”
“As artificial intelligence advances rapidly, we must work together in a bipartisan manner to proactively address the potential national security threats posed by this technology,” said Senator Hassan. “This bipartisan legislation establishes safeguards to ensure that the United States remains a world leader in responsible AI. We must continue to protect against potential safety risks by implementing balanced oversight of cutting-edge AI models to maintain our technological leadership while protecting public safety and the national interest. You can.”
background:
Artificial intelligence (AI) has the potential to dramatically improve and transform the way we live, but it also poses a variety of risks that could be harmful to Americans, including: Some can have devastating effects. Extremely powerful frontier AI could be exploited by foreign adversaries, terrorists, and less sophisticated bad actors to cause widespread harm and threaten U.S. national security. U.S. government, industry, and academic experts believe that advanced AI could one day enable or assist in the development of chemical, biological, radiological, nuclear, and cyberweapons. I am.
The Maintaining America’s Advantage in AI Act would establish an Artificial Intelligence Safety Review Office within the Department of Commerce. The agency is responsible for working with frontier Al companies, large data centers, and infrastructure-as-a-service providers (IaaS) to: (a) prevent exploitation of these industries by adversaries; (b) Ensure that advanced frontier Al models that pose chemical, biological, radiological, nuclear, and cyber risks undergo pre-implementation assessments similar to CFIUS.
Specifically, the bill:
Establishes the Office of Artificial Intelligence Safety Review, led by the Senate-approved Under Secretary of Commerce for Artificial Intelligence Safety. Instruct the roommates to do the following: Frontier Al Develop take-home tests for model developers to assess extreme risk of models before deployment. We work with the Al Safety Institute to provide best practices and technical assistance to the industry. Investigate future Al-related risks and report the findings to Congress. Requires state-of-the-art data centers that train state-of-the-art Al models to report to the Agency the ownership and location of their facilities. Protecting the industry from foreign adversaries by requiring LaaS providers to implement it. Customer recognition standards for transactions with foreigners. Protect Frontier Al models from cyber breaches and intellectual property theft by requiring developers to implement cybersecurity standards. Frontier Al helps model developers identify red teams. Best practices for testing chemical, biological, radioactive, nuclear, or cyber risks. Provides the Under Secretary a 90-day period to review the Frontier AL model’s safeguards against chemical, biological, radiological, nuclear, and cyber risks. Authorization to prevent unnecessary obstacles to implementation. Establish penalties for developers who do not comply with the law.
The full text of the Sustaining American Primacy in AL Act can be found here and a summary document can be found here.
