Sponsors (22)
The bill strengthens U.S. defenses, legal tools, and export controls to protect closed‑source AI models and IP through federal coordination, definitions, guidance, and listing/blocking authorities, but it does so at the cost of greater compliance burdens, concentrated executive power, potential diplomatic friction, and risks of chilling legitimate research and international collaboration.
Owners of U.S. closed‑source AI models (tech companies, model operators, and their workers) gain coordinated federal support — including definitions, detection/deterrence assistance, response help, an attackers list, and export restrictions — to protect models and IP from theft.
U.S. model owners and developers (companies and researchers) get clearer legal definitions of closed‑source models, 'model extraction,' and owner rights plus clearer export/reexport terms, reducing legal uncertainty and aiding IP enforcement and compliance.
Companies and researchers (especially those building or defending models) receive published best‑practice guidance, public consultations, routine risk assessments, and annual reporting to Congress — providing actionable defenses and ongoing governmental oversight of evolving threats.
Researchers, engineers, and international collaborators (tech‑workers, scientists, universities) could face a chilling effect because broad or vague definitions, aggressive countermeasures, and criminal/civil penalties may discourage legitimate research, interoperability testing, and cross‑border collaboration.
Small and medium firms, plus some larger companies (model owners and service providers), will likely absorb higher compliance, reporting, and legal costs — from consultations, information‑sharing, export restrictions, and potential law‑enforcement involvement — straining resources and operations.
Countries, governments, and taxpayers could face increased diplomatic friction and risk of escalation because public naming of alleged attackers, export restrictions, and punitive measures may provoke retaliation or harm foreign relations, especially if attribution is uncertain.
Based on analysis of 5 sections of legislative text.
Requires interagency assessments, public reporting, guidance, an attackers list, and authority to add model-extraction actors to the Commerce Entity List and to block property under IEEPA.
Official title: To prevent foreign adversaries from threatening the national security of the United States by extracting key technical features of closed-source, American-owned artificial intelligence models, and for other purposes.
Introduced April 15, 2026 by Bill Huizenga · Last progress April 15, 2026
Creates a federal effort to identify, report, and punish foreign actors that steal closed-source U.S. AI model capabilities through "model extraction" attacks. It requires interagency assessments, a public list of alleged attackers, guidance and information-sharing for model owners, and a process to add offending parties to the Commerce Department Entity List and to block or freeze property under IEEPA. Sets deadlines for agencies to complete an initial assessment and public report, publish best-practice guidance, establish a voluntary confidential reporting channel, and authorize export-control and sanctions-style actions against entities and fraudulent account providers that facilitate extraction attacks against U.S. closed-source AI models.