Artificial Intelligence Risk Evaluation Act of 2025
Sens. Hawley and Blumenthal Propose $1 Million Daily Fines for AI Firms Skipping Safety Tests
Legislative Progress
Key Points
Impact Analysis
Personal Impact
Small AI startups that happen to train models above the 10^26 operations threshold would face the same testing requirements as tech giants, potentially creating a significant barrier to entry. However, only the very largest models meet this threshold today, so most small businesses would be unaffected unless they are well-funded AI labs. The compliance costs — including delays for testing and the risk of $1M/day fines — could discourage smaller players from attempting frontier AI development.
Milestones
Read twice and referred to the Committee on Commerce, Science, and Transportation.
Sent to a congressional committee for expert review. The committee decides whether this bill moves forward.
Introduced in Senate
The bill was officially filed and given a number. It now enters the legislative queue.
Votes
No votes have been recorded for this legislation yet.
Related News
5 articlesExclusive: Hawley and Blumenthal unveil AI evaluation bill
Sens. Josh Hawley and Richard Blumenthal are introducing the Artificial Intelligence Risk Evaluation Act to establish a federal program within the Department of Energy to evaluate risks of advanced AI systems before deployment, targeting national security and 'loss-of-control' scenarios.
Senators propose federal approval framework for advanced AI systems going to market
The legislation mandates that AI models trained with computing power exceeding 10^26 floating-point operations undergo government testing. Companies failing to comply or releasing models without clearance would face fines of $1 million for every day the tool remains in operation.
Bill before Congress would push federal oversight into frontier AI
The Artificial Intelligence Risk Evaluation Act represents a shift toward mandatory federal oversight. It includes a controversial clause instructing the DOE to explore 'potential nationalization' if testing suggests the imminent rise of artificial superintelligence.
Source Information
Document Type
Congressional Bill
Official Title
Artificial Intelligence Risk Evaluation Act of 2025
Data Sources
Sponsor
Cosponsors
(2)Analysis generated by AI. Always verify with official sources.