The Federal Trade Commission (FTC) is seeking public comment on a proposed policy statement that would treat certain hidden changes to AI outputs as potential deception under Section 5 of the FTC Act.

 

What happened

Published in the Federal Register on July 7, 2026, the proposal says AI companies have marketed their systems as tools that aim to give users the best, most accurate response possible, so consumers may reasonably expect AI outputs to be truthful, objective, and aligned with their requests. The FTC says a company could mislead users if it secretly steers or suppresses model outputs for other objectives, including to satisfy state AI rules, without clearly disclosing that the system is prioritizing something different.

The proposal does not ban content moderation, safety limits, cybersecurity protections, or lawful product design choices; instead, it focuses on whether users are told the truth about how the system works and what aims guide its responses. The action follows a December 2025 executive order directing the FTC to clarify how federal deception law applies when state AI laws require changes to model outputs.

 

Going deeper

In the proposed policy statement, the FTC says, “Nonetheless, an AI company might be tempted to alter or steer the output of its systems contrary to consumers’ reasonable expectations for various reasons, including attempted compliance with a State law, such as Colorado’s recently revised Artificial Intelligence Act.” It goes on to state that companies could violate Section 5 if they secretly steer results for different objectives, even when doing so to comply with state AI laws. The notice points to Colorado’s revised automated decision-making law, which covers consequential decisions involving healthcare, employment, housing, lending, insurance, public benefits, and other high-impact services.

Patients and staff need to know when AI is operating under legal, safety, fairness, or institutional constraints that may affect its outputs. The concern aligns with a PLOS Digital Health study finding that “patients need specific information about an AI model, such as how it is overseen and how it will affect their health care.” Meanwhile, Paubox research shows why this issue is relevant operationally, as 89% of healthcare leaders call for AI for email threat detection, showing that AI is already becoming embedded in healthcare workflows before governance questions are fully settled.

 

Why it matters

The statement notes, “Of course, AI systems are likely to simultaneously pursue multiple objectives, some explicit and some implicit. This will often be consistent with consumers’ reasonable expectations.” For healthcare organizations, AI tools rarely serve only one purpose. The FTC acknowledges this reality, stating that AI systems are “likely to simultaneously pursue multiple objectives,” but its warning is that deception risk arises when the system is steered away from what users reasonably expect without clear disclosure.

Unfortunately, patients and providers may assume an AI tool is answering based on clinical relevance or the patient’s stated need when the system may also be prioritizing legal, operational, or institutional objectives. It makes transparency a practical compliance issue. As the study mentioned states previously, patients need to know how AI is used in the provision of their care.

See also: HIPAA Compliant Email: The Definitive Guide (2026 Update)

 

FAQs

Does the FTC regulate AI?

Yes, but not through one single AI law. The FTC governs AI mainly through its existing authority over unfair or deceptive business practices, unfair methods of competition, privacy, data security, advertising, endorsements, impersonation, and consumer protection.

 

What is AI washing?

AI washing is when a company uses AI hype to make a product, service, or business opportunity sound more advanced, effective, or profitable than it really is.

 

Does the FTC regulate AI impersonation and deepfakes?

The FTC has focused on AI-enabled impersonation because voice cloning, chatbots, and synthetic media can make scams more convincing.