WeVote
Log In

Bill

Bill

S 5105

Collaboration on Adversarial Threats and Security Risks Act

119th Congress Introduced by Jim Banks and 1 co-sponsor

The bill allows non-Federal entities to share information and coordinate to reduce AI security risks, with an antitrust exemption if done in good faith for security purposes.

Introduced in Senate
0
WeVote Research Nonpartisan
Bill Summary · S 5105

Overview

S. 5105, the Collaboration on Adversarial Threats and Security Risks Act, introduced July 23, 2026, aims to clarify and apply antitrust laws to the sharing and coordination around artificial intelligence (AI) frontier model risks. The bill focuses on collaboration among non-Federal entities to address covered AI security risks, while creating a guarded antitrust exemption under specific conditions.

Main purpose and intent

  • Establish that certain information sharing, assistance, and coordinated actions among non-Federal entities to prevent or mitigate AI security risks can occur without violating U.S. antitrust laws.
  • Enable collaboration specifically for reducing covered AI security risks, including actions such as delaying or limiting the deployment or development of AI, when done pursuant to defined procedures and for legitimate security purposes.
  • Provide a pathway for oversight and accountability through the Department of Justice’s Antitrust Division (Assistant Attorney General).

Key provisions and changes

  • Definitions (Section 2):
    • Antitrust Laws: Interpreted as per the Cybersecurity Information Sharing Act of 2015.
    • Artificial Intelligence, Assistance, Covered AI Security Purpose, and Covered AI Security Risk: Detailed definitions covering risk scenarios from theft, weaponization, and covert development to disruption of critical infrastructure, unauthorized access, and autonomous improvements that create security threats.
    • Unauthorized Access: Includes model weight/data theft, systematic querying to distill capabilities, and backdoors or data manipulation.
    • Non-Federal Entity and Exclusive Purpose: Clarifies who may participate and the scope of acceptable actions.
  • Antitrust Exemption (Section 3):
    • Exempts two or more non-Federal entities from antitrust liability when they provide or exchange information or assistance relating to a covered AI security risk in good faith for a covered AI security purpose.
    • Exemption also covers coordination or agreements to delay or limit AI deployment or development to reduce security risks, with prior written notice to the Assistant Attorney General detailing the risk and scope.
    • Limitations: Exemption does not permit price-fixing, market allocation, monopolistic actions, boycotts, or exchanging price information. The exemption is an affirmative defense; burden of proof lies with the entity.
    • Safeguards: Requires reasonable internal controls to ensure information is used only for security purposes; information submitted to the AG is confidential and exempt from public disclosure.
  • Injunctions (Section 4):
    • The Attorney General can seek injunctions against non-Federal entities to stop actions violating antitrust laws.
    • The act does not immunize activities if they fail to meet the good-faith and exclusive-purpose standard or if they increase overall AI security risks.

Who is affected

  • Non-Federal entities (e.g., private companies, universities, research labs) that collaborate on AI security risks.
  • The Department of Justice (Antitrust Division) and specifically the Assistant Attorney General who would receive notices and oversee coordination efforts.
  • The public and critical infrastructure sectors indirectly through the mitigation of AI security risks.

Procedural and timeline aspects

  • Introduction and referral to the Senate Judiciary Committee (July 23, 2026).
  • The bill outlines a notification process to the Attorney General prior to any coordinated delay or restriction, establishing procedural steps before actions occur.
  • The exemption operates as an affirmative defense in antitrust actions, requiring proof of good faith and exclusive security purpose.

This bill would, if enacted, balance collaboration among private and non-Federal actors to address AI security risks with carefully tailored antitrust protections and oversight.

Compiled from official sources — confirm details with the bill’s official record.

Sign in to ask a question.