AI Security and Innovation Act

Key Points

• The bill creates a new Center for AI Security and Innovation within the National Institute of Standards and Technology (NIST), focused on keeping the U.S. ahead in AI research, development, and evaluation while measuring national security and economic security risks.
• The center will evaluate AI security threats including model tampering, data leakage, adversarial prompt injection, model jailbreaks, and supply chain attacks. It will also assess whether AI systems could help create chemical, biological, radiological, or nuclear threats.
• Private companies can voluntarily share information with the center to test their frontier AI systems, and any information they share is exempt from public disclosure and cannot be used by any government entity to regulate that company.
• The bill authorizes $20 million per year for fiscal years 2027 through 2032 and allows the hiring of up to 15 highly paid technical experts. The center has a built-in sunset clause and expires after 5 years.
• The center explicitly has no regulatory, rulemaking, or enforcement authority. Its role is limited to research, evaluation, and publishing voluntary standards, guidelines, and best practices for AI systems.