Congress mandates AI testing before deployment—but keeps the data classified
S. 2938 — Artificial Intelligence Risk Evaluation Act of 2025 · Filed by Josh Hawley (R-MO) · 2 cosponsors · Introduced Sep 29, 2025 · Referred to committee
Your members of Congress
Enter a ZIP to see where your representative and both senators stood on this bill.
Looked up on this device — your ZIP is never stored on our servers.
What it does
This bill creates a mandatory federal testing program for advanced AI systems (those trained with more than 10^26 computing operations) run by the Department of Energy. Developers of such systems must participate, submit their code and training data for classified evaluation, and cannot deploy their AI publicly without compliance. The program will test for risks like loss of control, weaponization, and threats to civil liberties, then report findings to Congress to inform future AI regulation.
Why we flagged it
The bill's core function is to establish mandatory federal testing of advanced AI systems and generate empirical data for Congress to use in crafting permanent AI regulation. It is neither a ban nor a subsidy, but a procedural and informational mechanism designed to create a government-run evaluation bottleneck before deployment.
What the text implies
- The bill grants the Secretary of Energy (via DOE) sole authority to define what counts as 'advanced AI' and can unilaterally change that definition by rule, subject only to Congressional joint resolution—a high bar. This creates regulatory discretion that could narrow or broaden the scope of covered systems without explicit legislative action.
- Developers must surrender underlying code, training data, and model weights to a classified government program. This creates a centralized repository of proprietary AI intellectual property in federal custody, raising questions about data security, competitive disadvantage for smaller developers, and potential for government misuse or leaks.
The full analysis lists 5 implications of this text.
Who stands to gain
Department of Energy (expanded budget and authority); AI safety research contractors and consultants; Large AI developers with resources to comply (AMD, HPE, Cisco, Cognizant, Verisign)