S. 3952 — Future of Artificial Intelligence Innovation Act of 2026

This section establishes the Center for Artificial Intelligence Standards and Innovation within the National Institute of Standards and Technology (NIST) no later than 90 days after enactment. The Center's mission is to assist the private sector and federal agencies in developing voluntary best practices for the robust assessment of artificial intelligence (AI) systems, which may contribute to or inform standards development organizations. The Center must carry out the following functions, in coordination with NIST laboratories: (1) assessing AI systems using publicly available or voluntarily provided information and developing guidelines to improve secure AI development, deployment, and use; (2) supporting AI red-teaming (i.e., structured adversarial testing), sharing best practices, and coordinating testbeds with U.S. allies; (3) developing tools, methodologies, and guidelines to mitigate AI vulnerabilities such as data poisoning; (4) establishing AI blue-teaming (i.e., vulnerability evaluations and mitigation) capabilities and partnering with industry; (5) developing tools and guidelines for detecting synthetic content (i.e., AI-generated or modified media), including watermarking and provenance tracking; (6) coordinating metrics and methodologies for testing AI reliability, accuracy, and robustness; (7) coordinating with international partners for global AI standards interoperability; (8) producing resources for federal agencies to evaluate AI systems; (9) convening semiannual meetings with federal agencies and the private sector to share best practices and accelerate AI standards in sectors such as biotechnology, agriculture, and health care; and (10) examining safeguards and best practices to secure AI.

This section requires the Under Secretary of Commerce for Standards and Technology and the Secretary of Energy, in coordination with the Director of the National Science Foundation (NSF), to jointly establish not later than one year after enactment a testbed program to facilitate collaboration among Department of Energy (DOE) National Laboratories, other federal laboratories, the National Institute of Standards and Technology (NIST), the NSF National Artificial Intelligence Research Resource (NAIRR) pilot or successor, and public- and private-sector entities—including companies of all sizes—for testing, evaluating, and assessing security vulnerabilities of artificial intelligence (AI) systems; supporting related research and development (R&D); and developing standards to foster a third-party testing ecosystem. (As background, NAIRR provides researchers access to AI computational resources, data, and expertise.) In implementing the program, the Under Secretary and Secretary must (1) leverage existing advanced computing resources, testbeds, and expertise from DOE National Laboratories, federal laboratories, NIST, NSF, and private entities; (2) prioritize existing solutions and develop automated, reproducible tests where practicable, including those tailored to varying computational intensities; (3) assess and minimize computational needs for tests; (4) prioritize vulnerability assessments—including for autonomous cyber capabilities, cybersecurity, chemical/biological/radiological/nuclear threats, critical infrastructure, and energy security—using classified testbeds if needed and culminating in a hackathon; and (5) consider applicability to AI systems trained on biological sequence data that could enable pandemics or biological weapons. The section further directs the Under Secretary and Secretary to (1) develop metrics assessing the program's effectiveness in promoting collaboration, partnerships, public-private integration, and AI system use; (2) evaluate the program three years after its start using those metrics and report findings to specified congressional committees; (3) consult as appropriate with industry (including AI labs, companies, and financial sector representatives), academia, and civil society; (4) use any relevant preexisting program; and (5) establish a voluntary testing program for foundation model vendors and vendors of AI virtual agents or robots incorporating foundation models, across a range of modalities.

This section authorizes the Secretary of Commerce, acting through the Under Secretary of Commerce for Standards and Technology, and the Secretary of Energy to use the program established under section 102(b) as a testbed to identify, test, and synthesize new materials in order to advance materials science, energy storage and optimization, and advanced manufacturing for the U.S. economy through artificial intelligence, autonomous laboratories, and artificial intelligence integrated with emerging technologies such as quantum hybrid computing and robotics. The Secretaries must ensure that technologies accelerated under this testbed are supported by advanced algorithms and models, uncertainty quantification, software and workforce development tools to produce benchmark data, model comparison tools, and best practices guides. In implementing the testbed, the Secretaries shall enter into public-private partnerships, in consultation with industry, civil society, and academia, and may use resources from the Manufacturing USA Program, the Hollings Manufacturing Extension Partnership, National Laboratories, Federal laboratories, the private sector, and the section 102(b) program.

This section directs the Secretary of Commerce to coordinate activities under this subtitle to avoid duplication with Department of Energy (DOE) research entities—including National Laboratories and the Advanced Scientific Computing Research program—and relevant industries. It requires DOE or National Laboratory advanced computing resources, testbeds, expertise, or other resources provided to the National Science Foundation, National Institute of Standards and Technology, or other entities under this subtitle to be furnished on a reimbursable basis pursuant to a reimbursable agreement, which the Secretary may waive if necessary for Department of Commerce missions. It further specifies that nothing in this subtitle modifies requirements or authorities under section 5501 of the National Artificial Intelligence Initiative Act of 2020 (15 U.S.C. 9461)—DOE's cross-cutting artificial intelligence research and development program—or permits the Secretary of Commerce (including the Under Secretary of Commerce for Standards and Technology or Director of the Center for Artificial Intelligence Standards and Innovation) or the Director of the National Science Foundation to use DOE or National Laboratory monetary resources.

This section directs the Under Secretary of Commerce for Standards and Technology, in coordination with the Secretaries of Commerce and Energy, to submit to Congress, not later than one year after enactment, a report on implementation of sections 102 and 103. The report must include (1) a description of reimbursable agreements, statements of work, project schedules, and deliverables for the testbed program established under sections 102(b) and 103(a); and (2) details on the total amount of reimbursable agreements entered into under section 104(b); and may include such additional information as the Under Secretary determines appropriate.

This section directs the Under Secretary of Commerce for Standards and Technology and the Secretary of Energy to jointly lead federal agency information exchange and communication with the U.S. private sector and like-minded foreign governments on international technical standards for artificial intelligence. To support private sector-led engagement, it requires the Under Secretary to form alliances or coalitions with such governments (1) to promote AI standards development and adoption; (2) to encourage adoption of U.S. technical standards internationally and advocate for innovation-friendly AI governance countering foreign adversaries; (3) to facilitate collaboration on AI innovation, science, and R&D (including data sharing); (4) to develop supporting government-to-government infrastructure using existing agreements; (5) to identify cybersecurity best practices for AI models; and (6) to coordinate technology protection measures (e.g., in basic research and higher education) with the Secretary of State, National Security Council, and National Science Foundation. It further requires (1) technology trust criteria for participants ensuring high scientific and technological advancement and adherence to WTO Agreement on Technical Barriers to Trade principles (i.e., transparency, openness, consensus); (2) consultations among specified officials on AI innovation; and (3) participant commitments to intellectual property protections, safety standards, risk management, research security, and export controls for AI. It bars the People's Republic of China from any such alliance or coalition until the U.S. Trade Representative determines in the annual report required by 22 U.S.C. 6951 (i.e., assessing China's WTO accession commitments, including bilateral U.S. commitments) that China has achieved compliance; thereafter, the NIST Director and Secretaries of Commerce, Energy, and State must jointly report to Congress justifying China's participation, assessing national security risks (including technology transfer), addressing human rights concerns, evaluating monitoring of China's commitments, and providing related information. (Thus, China's exclusion links directly to ongoing U.S. monitoring of its WTO obligations.) It clarifies that nothing in the section prohibits persons (as defined in 5 U.S.C. 551) from specified activities.

This section directs the Comptroller General of the United States to submit to Congress, not later than one year after enactment of this Act, a report on regulatory impediments to innovation in artificial intelligence (AI) systems that includes the following: (1) significant examples of federal statutes and regulations directly affecting AI innovation and competition among companies of all sizes, accounting for voluntary standards and best practices developed with federal contributions; (2) an evaluation of progress in federal government adoption of AI and its use to improve government services; (3) an evaluation of, and examples where, AI assists federal agencies in delivering public services (including combating fraud) and ways to increase such use; (4) examples of federal laws and regulations on infrastructure and energy that unduly burden AI systems; and (5) recommendations for legislative or administrative action to increase the rate of AI innovation, based on the report's findings.