Reliable Artificial Intelligence Research Act of 2025

This bill establishes two federal prize competitions to advance research on two key AI safety areas: interpretability (making AI decisions understandable to humans) and adversarial robustness (resisting malicious attacks that cause harmful outputs). The Secretary of Homeland Security must launch these competitions within 270 days of enactment, consulting with agencies like the National Institute of Standards and Technology and AI industry experts. Competitions will evaluate submissions based on practical value in high-risk AI applications, such as healthcare or finance, and require participants to submit frameworks or models. The Secretary must report to Congress 180 days after the competitions end, assessing research gaps and suggesting policy actions. The bill authorizes $10 million in funding for these competitions over fiscal years 2026-2030.