Deterring American AI Model Theft Act of 2026

Protection of Intellectual Property and Economic Interests: The bill aims to protect the intellectual property rights and economic competitiveness of United States companies whose AI models are targeted by unauthorized acquisition of capabilities (Section 2).

National Security Safeguards: By deterring model extraction attacks, the bill seeks to prevent foreign adversaries from using "short cuts" to acquire advanced AI capabilities, thereby protecting national security and foreign policy interests (Section 2).

Support for Legitimate Research: The Act explicitly clarifies that authorized model training practices that follow terms of service remain a legitimate research method, ensuring that research is not unfairly targeted (Section 2, Section 3).

Information Sharing for Defense: The bill establishes an information-sharing mechanism so that AI model owners can voluntarily and confidentially share information about attacks with the Department of Commerce to improve defense (Section 4).

Public Awareness of Threats: The creation of the "AI Model Extraction Attackers List" and the publication of best practices provide transparency regarding specific entities and methods used in attacks (Section 4).

Economic Restrictions through Sanctions: The President has the authority to block and prohibit all transactions involving property and interests of identified entities of concern, which could impact various economic activities (Section 5).

Potential for Entity Listing: Entities identified as conducting model extraction attacks or using fraudulent networks may be added to the Department of Commerce's Entity List (Section 5).

Broad Definitions of Entities of Concern: The definition of "entity of concern" includes any foreign person or entity operating under the direction or control of a country of concern, which may impact a wide range of foreign actors (Section 3).