Sponsors (6)
The bill strengthens tools for U.S. model owners and the federal government to detect, deter, and block large‑scale model extraction—improving IP protection and national security—but does so by expanding executive authority, imposing compliance costs, and reducing transparency and some international collaboration.
U.S. companies that own closed‑source AI models can block access and freeze U.S. property/transactions of entities that performed or enabled large‑scale model extraction, reducing the ability of bad actors to monetize or deploy stolen models.
Model owners gain clearer federal definitions, policy framing, and export-control tools targeted at 'countries/persons of concern', helping Commerce and export reviewers enforce access controls and deter adversaries from acquiring advanced U.S. models.
Interagency assessments, a public best‑practices report, and a voluntary confidential information‑sharing mechanism give small businesses and tech teams actionable guidance and faster coordination to detect, respond to, and mitigate model‑extraction attacks.
Concentrating broad blocking authority (Entity List additions and IEEPA‑based restrictions) in the executive branch gives the administration major power over foreign entities and transactions with limited judicial review, raising due‑process and foreign‑relations risks.
New reporting, assessment, geo‑controls, monitoring, and enforcement expectations will impose material compliance and administrative costs on companies, banks, and taxpayers.
Labeling closed‑source models as essential national‑security assets may prioritize secrecy over transparency and independent auditability, limiting public oversight and scientific scrutiny of powerful AI systems.
Based on analysis of 5 sections of legislative text.
Directs Commerce to identify foreign actors and account networks stealing U.S. closed‑source AI models, report to Congress, and add offenders to the BIS Entity List with IEEPA blocking sanctions.
Official title: Prevent foreign adversaries from threatening the national security of the United States by extracting key technical features of closed-source, United States-owned artificial intelligence models, and for other purposes.
Introduced August 5, 2026 by William Francis Hagerty · Last progress August 5, 2026
Requires the Commerce Department to identify foreign actors and account networks that try to steal or copy proprietary U.S. AI models, report findings to Congress, and place offenders on the BIS Entity List with accompanying IEEPA blocking sanctions. It defines ‘‘closed‑source AI models,’’ ‘‘model extraction attacks,’’ ‘‘persons of concern,’’ and fraudulent account providers, sets deadlines for assessments and reports, and authorizes penalties and limited humanitarian and national‑security exceptions. The law directs interagency coordination, public consultations with model owners and experts, and a timeline (180 days for the initial assessment; reports and Entity List actions within about 210 days), plus follow‑up reporting and potential presidential waivers subject to congressional notice.