Bill · S. 3336· Senate(119-S-3336)
Reliable Artificial Intelligence Research Act of 2025
- Introduced
- 12/3/2025
- Last activity
- 12/3/2025 — Read twice and referred to the Committee on Homeland Security and Governmental Affairs.
- Sponsored by
- Maggie Hassan
- Cosponsors
- - 0 - 1
- Introduced
- In Committee
What this bill would do
The Reliable Artificial Intelligence Research Act of 2025 requires the Secretary of Homeland Security to conduct prize competitions aimed at advancing artificial intelligence research. One competition focuses on advancing the science of interpretability for commercially available or widely used artificial intelligence products. A second required competition focuses on developing artificial intelligence models that exhibit adversarial robustness for high-impact, high-risk applications. The Secretary must consult with various government officials and industry experts when organizing these competitions. Additionally, the Secretary is required to submit a report to congressional committees regarding the results and identified research gaps. The bill authorizes $10,000,000 for these purposes from fiscal years 2026 through 2030.
Yes· 3 of 7
Does this put more money into scientific research?+6
The bill provides $10,000,000 for research competitions related to artificial intelligence science.
Does this better protect the country from computer attacks?+5
The research into adversarial robustness specifically targets the ability of AI models to resist attacks that induce harmful outputs.
Does this give people more protection from unfair business practices?+3
The emphasis on interpretability and adversarial robustness aims to reduce risks of incorrect or harmful AI outputs for users.
The opposite
No question was scored the opposite.
Who it would affect, and how
Potential Positive Impacts for People
-
Increased AI Reliability: The bill seeks to advance interpretability, which helps humans understand how models make decisions (Section 3(a)). This could lead to more reliable artificial intelligence in high-risk, high-value use cases (Section 3(c)(3)(B)).
-
Improved Security Against Attacks: The requirement for adversarial robustness research aims to help models resist attacks that cause incorrect, restricted, or harmful outputs (Section 2(1), Section 4(a)). This could reduce the risk of adversarial attacks in high-risk, high-value applications (Section 4(c)(3)(B)).
-
Standard Creation: The research may create standards for artificial intelligence interpretability in both the government and industry (Section 3(c)(3)(C)).
Potential Negative Impacts for People
- No negative impacts are explicitly stated in the provided text.
Groups and Interests Ranking
Every group the model scored this bill for, from -10 to +10. Hover a score for the reason it gave.
- Economy And Finance
- Improves Consumer Protection+3Increases Job Creation+2Improves Economic Growth+1
- Education And Technology
- Increases Scientific Research Funding+6
- National Security And Foreign Policy
- Improves Cybersecurity Defense+5
- Demographic And Interest Groups
- Good For Business Owners+2
- Governance And Ideology
- Improves Government Transparency+2
- 12/3/2025
- Read twice and referred to the Committee on Homeland Security and Governmental Affairs.
- 1 Cosponsor added.
- Introduced in Senate
One bill Congress.gov lists as related to this one — companions in the other chamber, earlier versions, and bills this one absorbed or was folded into.
Jim BanksRepublican · Indiana