No Adversarial AI Act
- Bill Number
- S. 2177
- Origin Chamber
- Senate
- Congress
- 119th Congress, Session 1
- Policy Area
- Government Operations and Politics
- Status
- Introduced
- Latest Action
- 2025-06-25: Read twice and referred to the Committee on Homeland Security and Governmental Affairs.
- Last Updated
- 2025-12-05T22:55:33Z
AI-Generated Summary
Purpose
The "No Adversarial AI Act" aims to safeguard the U.S. government from risks posed by artificial intelligence (AI) developed or produced by foreign adversaries. It establishes processes to identify such AI and prohibits its acquisition or use by federal executive agencies, while allowing limited exceptions for critical needs.
Key Provisions
- Short Title: The legislation is titled the "No Adversarial AI Act."
- Development and Publication of a List of Foreign Adversary AI (Section 2):
- Within 60 days of enactment, the Federal Acquisition Security Council (FASC, a body that advises on securing government acquisitions) must create a list of AI produced or developed by a "foreign adversary" (defined as a covered nation like China, Russia, Iran, or North Korea, per existing law).
- Within 180 days, the Director of the Office of Management and Budget (OMB) must publish this list on a public website, in coordination with FASC.
- The list must be updated at least every 180 days.
- AI can be removed from the list if its owner provides a certification (with supporting evidence) that it is not linked to a foreign adversary, and FASC verifies this after review.
- Prohibition on Acquisition and Use (Section 3):
- Within 90 days of enactment, heads of executive agencies (federal departments and agencies like the Department of Defense or Homeland Security) must review their AI systems and exclude or remove any from entities on the list, working with FASC.
- Agencies must use existing legal authorities (under 41 U.S.C. § 4713, which covers safeguards against risks in federal contracting) to mitigate these risks.
- Exceptions: Agencies can approve limited use of prohibited AI if they notify OMB and relevant congressional committees in writing, and determine it is necessary for:
- Scientifically valid research (basic, unbiased studies as defined in education law).
- Evaluation, training, testing, or analysis.
- Counterterrorism or counterintelligence activities.
- Avoiding harm to essential government functions (e.g., national security operations).
- Definitions:
- Artificial Intelligence: Refers to systems that perform tasks like learning or decision-making, as defined in the National Artificial Intelligence Initiative Act of 2020 and related defense laws.
- Foreign Adversary Entity: Includes foreign adversaries themselves, foreign persons or companies based in or controlled by those nations, entities with at least 20% ownership by such persons, or anyone under their direction/control.
Significant Changes to Existing Law
- This bill builds on current federal acquisition rules (e.g., 41 U.S.C. § 4713 for risk mitigation in contracts) by introducing a targeted, regularly updated list specific to AI from foreign adversaries.
- It mandates proactive reviews and removals within tight timelines (60–180 days), which go beyond general supply chain security measures.
- No outright ban exists previously for AI specifically; this creates a new framework focused on AI, with a certification process for delisting that did not previously apply to this technology.
Potential Impacts
- On Government Agencies: Requires quick audits and potential replacement of AI tools, which could increase costs and administrative burdens but enhance cybersecurity. Exceptions may minimize disruptions for vital operations.
- On Citizens: Indirect benefits through stronger protection of government data and systems from foreign espionage or sabotage, potentially improving national security without direct effects on public AI access.
- On International Relations: Could strain ties with adversary nations by restricting U.S. government engagement with their AI technologies, signaling a push for domestic or allied alternatives amid global tech rivalries.
Main Stakeholders Affected
- Federal Executive Agencies: Must comply with reviews, removals, and exception processes, impacting operations in areas like defense, intelligence, and research.
- Federal Acquisition Security Council (FASC) and OMB: Responsible for list development, updates, and oversight, increasing their workload.
- Congressional Committees: The Senate Committee on Homeland Security and Governmental Affairs and the House Committee on Oversight and Accountability receive notifications for exceptions, gaining more visibility into agency decisions.
- AI Developers and Vendors: U.S. and foreign companies face scrutiny; those linked to adversaries may lose government contracts, while others could benefit from certification opportunities.
- National Security and Research Communities: Gain tools for risk mitigation but may need to navigate exceptions for legitimate uses like counterintelligence.
Notable Legal, Constitutional, or Political Implications
- Legal: Reinforces federal contracting laws by applying them specifically to AI, potentially setting precedents for regulating emerging technologies in supply chains. The certification and exception processes provide due process for affected entities, reducing arbitrary exclusions.
- Constitutional: Aligns with Congress's powers over federal spending and national defense (under Article I), prioritizing security without infringing on free speech or commerce rights, as it targets only government use.
- Political: Introduced bipartisansely (by Senators Scott and Peters), it reflects growing concerns over AI in U.S.-China tech competition. If enacted, it could influence broader AI policy debates, emphasizing security over innovation in federal contexts, and prompt similar state or international measures.
This summary was generated by AI and may contain inaccuracies. Refer to the official source document for the authoritative text.
Sponsor
Cosponsors (1)
Recent Actions
- 2025-06-25: Read twice and referred to the Committee on Homeland Security and Governmental Affairs.
- 2025-06-25: Introduced in Senate
Bill Versions
- No Adversarial AI Act — issued 2025-06-25 — PDF (6 pages)