H.R. 10594 — Reliable Artificial Intelligence Research Act of 2026

This section defines terms for the Act. It establishes adversarial robustness as the degree to which an AI model resists attacks that induce incorrect, restricted, or harmful outputs while preserving integrity and privacy; interpretability as the degree to which humans understand how an AI model makes decisions and responds to input changes; red-teaming as a structured adversarial process to identify vulnerabilities in AI systems; and the Secretary as the Secretary of Homeland Security. It adopts the definition of artificial intelligence from section 5002 of the National Artificial Intelligence Initiative Act of 2020.

This section requires the Secretary of Homeland Security to commence at least one prize competition within 270 days of enactment to advance artificial intelligence interpretability for commercially available or widely used AI products. The Secretary must consult with the Secretary of Commerce, the NIST Director, the National Cyber Director, the NSF Director, and relevant AI industry experts, and may contract with public or private entities to administer the competition.

This section requires the Secretary to commence at least one prize competition within 270 days of enactment under the Stevenson-Wydler Technology Innovation Act to develop artificial intelligence models exhibiting adversarial robustness for high-impact, high-risk government or industry applications. It directs consultation with the Secretaries of Commerce and Energy, the National Cyber Director, the Directors of NIST and the National Science Foundation, and relevant agency heads and industry experts, and permits the Secretary to structure the competition in phases and to contract with public or private entities for administration.

This section requires the Secretary to submit a report to the Senate Committee on Homeland Security and Governmental Affairs and the House Committee on Homeland Security within 180 days after the first prize competition concludes. The report must evaluate how competition results inform the fields of interpretability and adversarial robustness, assess any gaps identified during the competitions, and recommend congressional actions to advance those fields.