HR 9914
To establish the applicability of antitrust laws to the sharing of artificial intelligence frontier model risks, and for other purposes.
TL;DR
Rep. Robert Latta (R-OH) introduced this bill to give AI companies an antitrust safe harbor when they share information about safety risks in frontier AI models with each other and the government. In plain English: if OpenAI, Google, and Anthropic want to compare notes on dangerous AI capabilities, this bill says the DOJ can't sue them for collusion. It's designed to encourage cooperation on AI safety without triggering antitrust liability.
How This Might Impact Your Business
Frontier AI developers (OpenAI, Anthropic, Google DeepMind, Meta AI, and similar large model builders) get legal cover to share safety information, capability risks, and threat data with competitors.
Companies building on top of frontier models (enterprise SaaS, AI-powered healthcare tools, fintech using LLMs) may benefit indirectly from better industry-wide safety standards, but face no new compliance burden.
No new reporting requirements, fees, or penalties are created for any business.
Smaller AI startups and non-frontier model developers are largely unaffected since the bill targets 'frontier' models (typically defined as the largest, most capable systems).
Antitrust exposure decreases for participants in AI safety consortiums like the Frontier Model Forum or MLCommons.
No implementation deadline or effective date pressure since this is permissive, not mandatory.
Downstream effect: expect more industry-wide safety disclosures and shared risk frameworks, which could shape voluntary standards your vendors adopt.
What Should You Do
Ask your legal team whether your AI vendors participate in frontier model safety-sharing groups, and whether their disclosures could affect your product roadmap.
If you build or deploy frontier-scale models, have counsel review your participation in industry safety forums; this bill could expand what you can legally discuss with competitors.
Track the bill's progress through the House Judiciary Committee; companion Senate action would signal real momentum.
Monitor whether shared safety information from competitors reveals risks in AI systems you already use, and update your AI risk register accordingly.
No immediate compliance work required, but flag this to your government affairs team as a signal of the direction of federal AI policy.
Who It Affects
Sponsors
Status Timeline
committee
Referred to the House Committee on the Judiciary.
July 23, 2026