Deterring American AI Model Theft Act of 2026
Current understanding
HR8283, the Deterring American AI Model Theft Act of 2026, advances a dedicated federal framework aimed at preventing theft of U.S.-developed AI models, treating trained model weights and related artifacts as national-security-sensitive assets. The bill was ordered reported amended by a unanimous 43-0 committee vote, signaling bipartisan committee-level support though full chamber consideration and enacted text remain pending. Practical and fiscal implications (agency roles, penalty structures, and any authorized appropriations) will be clarified as the reported substitute text becomes available.
Evidence log
- 2026-04-22 — Deterring American AI Model Theft Act of 2026: cross-connection with advanced-ai-innovation-security-eo: Both address federal AI security posture — the EO on innovation/security policy and HR8283 on protection of trained model IP from theft. (novelty: 3)