AI Security and Innovation Act

This bill amends the National Artificial Intelligence Initiative Act of 2020 to establish a Center for AI Security and Innovation within the National Institute of Standards and Technology (NIST) to ensure continued U.S. leadership in AI research, development, and evaluation, and to address AI-related security risks.

Sec. 2, Artificial Intelligence Advancement and Reliability: This section adds new definitions to the Act, including 'artificial intelligence model' and 'artificial intelligence system', and defines 'foreign adversary' and 'intelligence community'. It establishes the Center for AI Security and Innovation within NIST, to be led by a Director appointed by the Secretary of Commerce. The Center's duties include measuring AI system risks (including national and economic security), supporting information exchange for risk mitigation, evaluating and improving AI security measures (such as against model tampering, data leakage, and supply chain attacks), establishing voluntary agreements with entities to evaluate 'covered frontier systems', conducting evaluations and assessments, supporting NIST labs in developing standards and best practices, assessing trends in AI development in the U.S. and foreign adversaries, and consulting with various federal agencies and AI developers. The Director must define 'covered frontier system' within 180 days. The Center may hire up to 15 critical technical experts with special pay authority. Information shared with the Director is confidential and cannot be made public without consent or used for regulation. The Director must avoid duplication with other NIST activities, may engage internationally (except with foreign adversaries), and must submit annual reports to Congress starting fiscal year 2027. The Center has no regulatory or enforcement authority. It is authorized $20 million per year for fiscal years 2027 through 2032, and terminates 5 years after enactment.