S-5105-119
Read twice and referred to the Committee on the Judiciary.
Sponsored by Adam Schiff (D-CA)
What it does
This bill would create a limited exemption from federal antitrust laws for competing AI companies that share information or coordinate actions specifically to address serious AI security risks — such as AI being weaponized by foreign adversaries, used to develop weapons of mass destruction, or used to attack critical infrastructure. Companies that want to coordinate on delaying or restricting AI releases for security reasons would be required to notify the Justice Department's Antitrust Division in advance. The exemption would not cover price-fixing, market allocation, or other standard antitrust violations, and companies claiming the exemption bear the burden of proving their actions were taken in good faith for security purposes only.
Who benefits
Large AI developers (such as frontier model companies) who could share threat intelligence and coordinate safety responses without antitrust liability. National security agencies and the broader public who would benefit from faster, coordinated responses to AI-related threats. Critical infrastructure operators who could benefit from AI companies jointly identifying and mitigating risks to power grids, water systems, and financial networks. Smaller AI companies that lack the resources to independently identify all security risks and could benefit from shared threat information. The Justice Department's Antitrust Division, which gains a formal notification and oversight role.
Who is hurt
Competitors and new market entrants who could be disadvantaged if large incumbents use coordinated "security" delays to slow the release of rival AI products. Consumers and businesses that rely on timely AI product releases, who could face delays if coordination is used beyond its intended scope. Plaintiffs in private antitrust actions, who retain their rights but may face a harder burden if defendants invoke the security exemption. Open-source AI developers and researchers who may be subject to coordinated restrictions on model releases. Journalists and the public, who would be denied access to DOJ notification filings under a mandatory FOIA exemption.
Supporters argue
Supporters argue that AI frontier models pose genuinely novel national security risks — including potential use by adversary nations like China and Russia to develop biological or cyberweapons — and that competing companies currently cannot legally coordinate to address these threats without risking antitrust liability. They contend the bill is narrowly tailored, mirroring the structure of the Cybersecurity Information Sharing Act of 2015, which successfully enabled cybersecurity threat-sharing without enabling anticompetitive abuse. The mandatory DOJ notification requirement and the explicit exclusion of price-fixing and market allocation provide meaningful guardrails against misuse.
Opponents argue
Opponents argue that the bill's broad definition of "covered artificial intelligence security risk" — which includes vague categories like AI that "substantially reduces" human oversight — could be invoked pretextually by dominant companies to delay competitors' products under the guise of safety. They contend that the mandatory FOIA exemption for DOJ notifications eliminates public accountability, making it nearly impossible for courts, regulators, or the public to detect anticompetitive abuse. Critics further argue that the AI industry already has significant market concentration, and that antitrust exemptions historically have been difficult to cabin once enacted, citing the insurance industry's McCarran-Ferguson Act as a cautionary example.