AI Security and Innovation Act

National and Economic Security: The Center is tasked with measuring risks related to artificial intelligence systems, specifically focusing on national security and economic security risks (Section 5304(a)(2)(A)).

Risk Mitigation through Information Exchange: The bill supports the exchange of information between non-governmental entities and Federal agencies to help mitigate identified risks (Section s5304(a)(2)(B)).

Improved AI Security Measures: The Director is responsible for evaluating and improving security measures to reduce the risk of misuse, specifically targeting threats like model extraction, data leakage, and supply chain attacks (Section 5304(c)(1)(A)).

Standardized Safety Guidelines: The Center will support the development and voluntary adoption of standards, guidelines, and best practices regarding the testing, evaluation, and reliability of artificial intelligence systems (Section 5304(c)(1)(D)).

Protection of Private Information: Any information shared by a covered entity or developer with the Director is exempt from certain public disclosures and cannot be used by any level of government to regulate the activity of that entity related to the information (Section 5304(e)).

Risk of Misuse: The bill explicitly notes the need to address risks of misuse of artificial intelligence systems, including threats such as model tampering and adversarial prompt injection (Section 5304(c)(1)(A)).

Potential Security Vulnerabilities: The Center is tasked with assessing potential security vulnerabilities, flaws, or malign foreign activity that may result from artificial intelligence systems (Section 5304(c)(1)(C)(ii)).