HR-9363-119
Ordered to be Reported in the Nature of a Substitute (Amended) by the Yeas and Nays: 29 - 0.
Sponsored by Jay Obernolte (R-CA)
What it does
This bill would amend the National Artificial Intelligence Initiative Act of 2020 to establish a Center for AI Security and Innovation within the National Institute of Standards and Technology. The Center would evaluate security risks of advanced ("frontier") AI systems, develop voluntary standards and best practices, conduct classified and unclassified risk assessments, and coordinate with other federal agencies and international partners; it would receive $20 million annually for fiscal years 2027-2032 and terminate after five years unless renewed.
Who benefits
AI developers and companies (covered entities) that participate in voluntary evaluations and gain confidentiality protections for shared information; federal agencies (Commerce, Defense, Homeland Security, intelligence community) that gain a coordinating body for AI security assessments; national security officials seeking centralized comparative assessments of U.S. versus foreign adversary AI capabilities; NIST, which gains new staff and funding authority.
Who is hurt
State, local, and Tribal governments, which are explicitly barred from using information shared with the Center to regulate covered entities, potentially limiting their independent oversight of AI companies; taxpayers funding the $20 million annual appropriation; the public, which may have limited access to evaluation findings since publication is left to the Director's discretion; smaller AI developers without resources to participate in voluntary evaluations, who may face a perceived disadvantage relative to larger firms that can engage with the Center.
Supporters argue
Supporters argue that as AI systems advance rapidly, the federal government needs a dedicated, technically capable body to assess national security risks like model jailbreaks, data leakage, and supply chain attacks before adversaries or bad actors exploit them. They contend the voluntary, non-regulatory structure—paired with confidentiality protections for shared data—encourages industry cooperation that mandatory reporting requirements might discourage, while still allowing the U.S. to track competitiveness against foreign adversaries in AI capabilities.
Opponents argue
Opponents argue that exempting information shared with the Center from public disclosure and barring any state, local, or Tribal use of that information for regulation could shield AI companies from accountability and undercut state-level AI oversight efforts already underway. They contend that a voluntary, non-regulatory framework with discretionary publication of findings risks becoming a mechanism for industry-friendly information gathering rather than meaningful public risk disclosure, especially given the Center's five-year sunset and limited enforcement teeth.