This bill, titled the Artificial Intelligence Risk Evaluation Act of 2025, mandates the Secretary of Energy to establish an Advanced Artificial Intelligence Evaluation Program within 90 days of enactment. Congress believes rapidly advancing artificial intelligence (AI) capabilities present significant risks to national security, public safety, and civil liberties, necessitating a secure testing program. The program's primary purpose is to provide Congress with empirical data and insights for effective Federal oversight of AI, ensuring regulatory decisions are based on thorough testing to safeguard American citizens.

The program will conduct standardized and classified testing of advanced AI systems, defined as those trained using a quantity of computing power greater than 10^26 operations, to systematically collect data on the likelihood of adverse AI incidents. These incidents include loss-of-control scenarios, risks of weaponization, threats to critical infrastructure, and scheming behavior. Developers of these advanced AI systems are obligated to participate and provide necessary materials, with non-compliance incurring a penalty of not less than $1,000,000 per day. The program will also develop recommended containment protocols, contingency planning, and mitigation strategies based on testing data, and inform the creation of evidence-based standards and regulatory options.

Furthermore, the program is tasked with assisting Congress in determining the potential for AI systems to reach artificial superintelligence, exceed human oversight, or pose existential threats. Within 360 days, the Secretary must submit a detailed recommendation to Congress for a permanent Federal oversight framework, drawing directly from the program's insights and data. This plan will summarize testing outcomes, recommend evidence-based standards, certification procedures, and licensing requirements, and propose adaptive governance strategies. The program is set to terminate seven years after enactment unless renewed by Congress.