Sponsors (4)
The bill increases transparency and oversight of foundation models—helping consumers, high‑risk sector buyers, and smaller developers make safer choices—while imposing new compliance costs, raising privacy and competitive‑disclosure risks, and expanding regulatory reach.
Users, downstream developers, and buyers of AI (e.g., tech workers, students, small-business owners) will get clearer information because developers must disclose training data sources, model capabilities, limitations, and safety measures, improving transparency and enabling more informed use and evaluation of models.
Consumers and organizations (taxpayers, students, hospitals/health systems) gain easier access to standardized model information through a consumer‑friendly website and an FTC-hosted, machine‑readable central repository, making comparison, procurement, and oversight simpler.
Small and new businesses will face lower short‑term burdens because the bill provides tailored compliance help (grace periods, technical assistance, and templates) to ease meeting disclosure requirements.
Covered entities (tech firms, startups) will face new compliance costs for data collection, documentation, and reporting that could raise product prices, divert resources from innovation, or slow deployment.
Public disclosure requirements risk revealing proprietary model details or sensitive information (even with redaction rules), which could harm commercial competitiveness or expose security vulnerabilities.
Mandating disclosures about inference‑time data collection and retention could create privacy risks for end users and increase compliance complexity if user data protections are insufficiently specified.
Based on analysis of 2 sections of legislative text.
Requires the FTC to promulgate rules forcing foundation-model developers/providers to submit and publicly disclose standardized model metadata, data provenance, governance, performance, and safety information.
Official title: To direct the Federal Trade Commission to establish requirements for making information available to the public about the training data and algorithms used in artificial intelligence foundation models, and for other purposes.
Introduced March 26, 2026 by Donald Sternoff Beyer · Last progress March 26, 2026
Requires the Federal Trade Commission to write rules within one year that force companies who develop or provide large "foundation" AI models to disclose standardized information about those models. The rules will require submission of model metadata to the FTC, public posting (human- and machine-readable) of specified details such as training data sources, data governance, capabilities, limitations, evaluations, versioning, and incident monitoring, while allowing limited redactions for sensitive data and permitting disclosures inside system/model cards.