Selling software or services that include AI in United States (federal): related rules
Organizations that build and sell AI-enabled products or services to businesses or consumers. They take on provider duties and, above all, customer demands for evidence of governance, security and transparency before buying.
Applies if you sell or promote products or services described as "AI-powered" or your advertising claims AI capabilities. Using AI to create marketing content does not by itself trigger this obligation unless the ad makes claims about AI.
What the rule provides
Claims about AI accuracy, results, savings or autonomy must be backed by reasonable evidence before they are published.
Do not say a product uses AI if it does not, and do not overstate what it does.
Review sales and marketing material (website, ads, client proposals) against written criteria.
Typical evidence
Evidence file for each claim (tests, metrics, date, model version)
Pending confirmation in the official source (last checked: November 4, 2026)The legal basis (15 U.S.C. § 45(a)) is checked on govinfo. FTC cases on AI claims and their dates come from secondary sources; no official text sets which evidence suffices in each case, so it is presented as enforcement practice.
Base documents on topics that procurement questionnaires from banks and US clients often cover: policy, inventory, impact assessment, incidents, vendors and security.
Six base documents for integrating third-party models through APIs or RAG: due diligence, model provenance, architecture, integration map, adversarial security and service level agreement.
Seven base documents for hosting in-house or open-weight models: provenance, security baseline, adversarial testing, architecture, model card, life cycle and rollback.