S 5105
A bill to establish the applicability of antitrust laws to the sharing of artificial intelligence frontier model risks, and for other purposes.
TL;DR
Senator Adam Schiff (D-CA) introduced this bill to give AI companies legal cover to share information about safety risks in frontier AI models without triggering antitrust violations. It clarifies that competitors collaborating on AI risk assessment, things like security vulnerabilities or dangerous capabilities, would not automatically be considered illegal collusion.
How This Might Impact Your Business
Frontier AI developers (think OpenAI, Anthropic, Google DeepMind, Meta) would gain a safe harbor to share safety and risk information with competitors without antitrust exposure.
Only applies to sharing about AI model risks (safety flaws, dangerous capabilities, security vulnerabilities), not pricing, market allocation, or commercial terms.
Companies deploying frontier AI models could benefit from faster, industry-wide risk disclosures, potentially reducing the odds of using a model with an undisclosed safety issue.
No new compliance mandates, reporting requirements, or penalties are created for businesses.
Enterprise AI buyers may see vendors participate in new industry safety consortiums or information-sharing groups as a result.
Bill is at the earliest stage (referred to Senate Judiciary Committee) and has no companion House bill yet, so movement is uncertain.
Smaller AI developers and downstream deployers (SaaS, healthcare AI, fintech AI users) are not directly regulated.
What Should You Do
If you build or fine-tune frontier models, ask your legal team whether existing safety information-sharing arrangements could be expanded under this proposed safe harbor.
If you procure AI from major vendors, add a contract question about whether they participate in any AI safety information-sharing consortiums and how they would notify you of shared risk findings.
Monitor the Senate Judiciary Committee schedule for hearings; no markup has been announced yet.
Brief your risk and compliance teams that this bill does not create new obligations, so no immediate policy changes are needed.
Watch for related bills (like the AI safety disclosure proposals) that could pair with this one and create actual reporting duties.
Who It Affects
Sponsors
Status Timeline
committee
Read twice and referred to the Committee on the Judiciary.
July 23, 2026