Representatives Jay Obernolte (R-CA) and Lori Trahan (D-MA) released a discussion draft of the Great American Artificial Intelligence Act on 4 June 2026. The release is at the pre-introduction, public comment stage: the draft has not been formally introduced in the House of Representatives and the sponsors are soliciting technical and policy feedback before introduction. The proposal is bipartisan and co-sponsored by Representatives Scott Franklin (R-FL), Suhas Subramanyam (D-VA), Erin Houchin (R-IN), and Scott Peters (D-CA).
The draft designates the Center for AI Standards and Innovation, housed within the Department of Commerce, as the lead federal administrator for AI safety. CAISI would develop voluntary guidelines and best practices for the AI industry, test and evaluate frontier AI systems, and administer a mandatory third-party audit and verification regime. Developers of large frontier AI systems - defined by reference to compute thresholds not yet fixed in the draft - must publish and maintain documented plans to address catastrophic risk, report safety incidents to federal and state regulators, and submit to periodic third-party audits. Civil penalties and injunctions are the proposed remedies for non-compliance.
Frontier AI developers - companies training large language models and other high-capability AI systems at or above the draft's compute threshold - bear the core obligations under the proposal. The mandatory third-party audit requirement necessitates engagement with accredited assessors, and the incident reporting obligation creates a disclosure channel running to both federal and state regulators, which may trigger parallel state-level enforcement. The draft does not identify platform operators and enterprise deployers of third-party frontier models as primary obligation-holders.
All provisions remain subject to revision following the public comment period. The definition of 'frontier AI system' - specifically the compute threshold - is not finalised, nor are the definition of 'catastrophic risk' and the accreditation criteria for third-party auditors. The draft summary materials released to date do not address federal preemption of state AI laws, leaving open whether the Act would supersede existing obligations under statutes such as Colorado SB 205 and Connecticut Public Act 26-15.
Licentium monitors U.S. federal and state AI legislative developments and advises clients on positioning for emerging regulatory obligations. We assess how proposed federal AI measures interact with existing state-level requirements and advise on engagement strategies during public comment periods. Work we undertake includes frontier AI regulatory analysis, compliance strategy for compute-threshold obligations, third-party audit programme design, and multi-jurisdictional AI governance advisory.