The bill would require the Secretary of Homeland Security to start at least two prize competitions within 270 days of enactment. One would support research into making artificial intelligence decisions easier for people to understand. The other would develop AI models that can better resist attacks that could cause incorrect, restricted, or harmful outputs. The Secretary would consult with federal science, technology, and security officials and relevant industry experts. After the first competition ends, the Secretary would report to Congress on results, research gaps, and possible next steps.
The bill focuses on research competitions, not direct services to individuals. It does not specify changes to how people use AI products. It aims to support research on making AI more understandable and resistant to attacks, especially in high-risk uses.
The bill would authorize $10 million for the Secretary to carry out its provisions for fiscal years 2027 through 2031. This is an authorization, not a statement that the funding has been appropriated.
No publicly available information.
No publicly available information.