Values for America · Brief for congressional offices

To the office of U.S. Representative Gabe Evans

Re: HR 10152 (119th Congress), Open-Source AI Leadership Act · Latest action (2026-09-16): Ordered to be Reported by the Yeas and Nays: 26 - 21.
VFA SI Solution #445 · published 2026-10-17 · https://www.valuesforamerica.com/solutions/445
Office of record (Congress.gov): 1229 Longworth House Office Building, Washington, DC · (202) 225-5625

Limited Review and Open Model Support Act for Artificial Intelligence

The measure, as described from the official record: Help people and firms understand risks from open artificial intelligence models linked to China, Iran, North Korea, or Russia, and support wider use of open models built in the United States.

The proposal

Congress has power here, and that power has limits. Article I, Section 8, Clause 3 authorizes federal review of foreign commerce, which governs public risk review of open artificial intelligence models developed or made available by China, Iran, North Korea, or Russia. Article I, Section 8, Clause 1 authorizes capped spending for that work, and Article I, Section 8, Clause 8 supports voluntary help for domestic science and research. Those grants do not authorize bans on lawful models, punishment of users, or federal control of speech and research. The strongest objection is that risk reports plus Commerce support for United States models could chill speech and research by steering users away from disfavored models. The VFA proposal answers that objection directly. It bars bans, blocks on downloads, penalties based on model choice, and pressure on private hosts to remove lawful speech. It requires written risk factors, named sources, and a chance to correct factual errors before final publication, consistent with the First Amendment and Fifth Amendment Due Process, and it leaves state consumer protection and state research programs in place under the Tenth Amendment. The VFA proposal would keep the core goal of the Open-Source Artificial Intelligence Leadership Act and make it narrower, clearer, and limited in time. The Department of Commerce would review open artificial intelligence models linked to China, Iran, North Korea, or Russia and publish plain-language risk notes on loss of personal or business information, weak controls against misuse, and harm to economic or national security. Annual reports to Congress would compare those models with open models built by people and firms in the United States on cost, performance, and adoption. Reports would protect private information and would not disclose classified facts or the private records of users. The VFA proposal would change Commerce support for United States open models from broad action into defined voluntary help. Commerce could publish guides, testing tools, and plain-language comparisons. It could name barriers such as computing cost, poor records, and weak safety testing, then suggest low-cost fixes. The VFA proposal would control cost and protect state authority. Spending would be capped at $25 million per year for five years, paid from existing Commerce funds subject to annual appropriation. The program would end after five years unless Congress renews it after a public hearing. Nothing in the measure would displace state consumer protection law or state research programs, and courts would remain open to hear claims over speech, due process, or lawful limits on federal power.

Operative provisions (VFA proposal)

effective date: 180 days after enactment for the first risk review cycle
enforcement mechanism: Commerce publishes standards and risk reports; no ban on models, no block on downloads, and no penalty on private users for their model choices; Commerce support is limited to voluntary technical help and public information; judicial review remains open for final agency action on speech, due process, and lawful limits on federal power
funding amount: Capped at $25 million per year for five years, subject to annual appropriation
funding source: Department of Commerce salaries and expenses appropriation, with no new fee
oversight body: Department of Commerce, with annual reports to Congress and review by the Government Accountability Office
sunset years: 5 years

Constitutional basis

Article I, Section 8, Clause 3
Article I, Section 8, Clause 1
Article I, Section 8, Clause 8
First Amendment
Fifth Amendment Due Process
Tenth Amendment

Concerns with the measure as written:

Broad Commerce action to support domestic models could burden First Amendment protected speech and research if carried out through pressure on private publication, hosting, or distribution of lawful models.
Public naming of models and makers requires Fifth Amendment Due Process through clear written criteria, stated sources, and a chance to correct factual errors.
Federal work should preserve state consumer protection law and state research programs consistent with the Tenth Amendment.

Why both sides can back it (VFA’s assessment)

VFA assesses a gain for Republicans and conservatives in stronger review of foreign adversary models linked to China, Iran, North Korea, or Russia, protection of personal and proprietary information, and strict cost control through a capped budget, a five-year sunset, and review by the Government Accountability Office. VFA assesses a gain for Democrats and progressives in support for open research and small and public-interest users of United States open models, public plain-language risk information, voluntary testing tools and guides, and privacy safeguards with a chance to correct factual errors.

Strongest objection, and the answer: The strongest objection is that risk reports plus Commerce support for United States models could chill speech and research by steering users away from disfavored models. The VFA proposal answers that objection directly.

Check before relying on this

Basis: The official summary on Congress.gov for the Open-Source Artificial Intelligence Leadership Act states the Commerce duties to identify, assess, and publish risks of open models from China, Iran, North Korea, or Russia, the annual report to Congress comparing foreign and United States open models on adoption, cost, and performance, the Commerce authority to support adoption of United States open models, and committee action on 2026-09-16 by a vote of 26 to 21.

The full bill text and current law are not stated in the official summary; eligibility rules, existing appeal rights, cost effects, conflicts with current Commerce authority, and final legislative status should be verified at Congress.gov before action. Written standards, correction process, privacy protections, spending cap, and sunset are new VFA proposals.

SI-generated · passed automated checks (Constitution citation, source verification, independent SI critic) · not individually reviewed by a person · Report an error: hello@valuesforamerica.com

Values for America is a nonpartisan, constitution-first SI think tank. This brief is informational. It is a policy proposal, not law, and it does not support or oppose any candidate or party. Full analysis and sources: https://www.valuesforamerica.com/solutions/445 · Methodology: https://www.valuesforamerica.com/methodology