FTC AI Enforcement in 2026: What Businesses Actually Need to Know

What the FTC actually published in 2026

On July 1, 2026, the Federal Trade Commission announced that it was seeking public comment on a proposed policy statement addressing AI accuracy. The proposal concerns AI companies that may distort system outputs to achieve undisclosed ideological objectives in ways that conflict with reasonable consumer expectations about objectivity and accuracy.

The proposal discusses how that conduct could be unfair or deceptive under Section 5 of the FTC Act. It was not a final rule, and it did not create a blanket requirement that every business label every AI interaction, publish an explanation of every model, or comply within 90 days. The public-comment deadline stated by the FTC was July 31, 2026.

Different laws can impose additional obligations for particular systems and markets. For example, the EU AI Act guide explains the separate European risk and transparency framework.

What the FTC already enforces

The absence of the previously claimed mandate does not mean AI products are outside consumer-protection law. The FTC has used its existing authority against deceptive AI claims and AI-enabled schemes.

In September 2024, the agency announced Operation AI Comply, which included actions involving allegedly deceptive claims about AI-powered services and business opportunities. In February 2025, the FTC finalized an order involving DoNotPay over claims that its service could operate like a human lawyer without adequate evidence.

The practical lesson is straightforward: calling a product "AI-powered" does not relax the normal requirement that advertising claims be truthful, supported, and not misleading. A business should be able to substantiate claims about accuracy, performance, savings, legal compliance, safety, and other material capabilities before making them.

What businesses should review

1. Inventory public AI claims

List the AI-related claims made on landing pages, in sales materials, inside the product, in app-store listings, and through customer-support scripts. Record the evidence supporting each measurable claim and the conditions or limitations that affect it.

2. Remove or qualify unsupported claims

Do not claim that an AI system is more accurate, unbiased, compliant, secure, or capable than the available evidence shows. If a limitation would materially change how a reasonable customer understands the product, make the limitation clear near the claim instead of relying on a buried policy.

3. Make disclosures that prevent deception

There is no single FTC rule requiring the same AI label in every context. A disclosure may nevertheless be necessary when omitting it would make a representation misleading. The appropriate disclosure depends on what the system does, what the business claims, the audience, and other laws or platform rules that apply.

4. Match privacy statements to actual data practices

Document what information an AI feature receives, which service providers process it, whether information is retained or used for model improvement, and what controls are available to users. Privacy policies and in-product notices should describe the real implementation rather than generic AI language. The AI privacy disclosure guide provides an implementation checklist and examples.

5. Review high-impact use cases separately

AI used in employment, housing, credit, insurance, health, education, or other consequential decisions may trigger additional federal, state, sector-specific, or platform obligations. An FTC-focused review is not a substitute for assessing those requirements.

Claims this article does not make

• There was no verified March 11, 2026 FTC AI policy statement matching the earlier description.
• The July proposal did not impose a universal AI-disclosure checklist on every business.
• The FTC announcement did not establish the previously claimed 90-day grace period.
• There is no automatic $50,120-per-violation-per-day penalty schedule for the supposed requirements described in the earlier article.

A defensible compliance workflow

1. Map each AI feature, provider, data flow, user group, and decision it influences.
2. Collect every public capability, performance, accuracy, and compliance claim.
3. Link each material claim to evidence and an internal owner.
4. Add specific disclosures where they are needed to keep a claim or interaction from being misleading.
5. Align the privacy policy, terms, app-store disclosures, and in-product notices with the implementation.
6. Re-review the product when a model, provider, data use, or material capability changes.

The PolicyForge AI policy generator can help organize AI-related policy language, but generated text should be checked against the product's real behavior and the laws that apply to the particular use case. For higher-risk decisions or unresolved legal questions, obtain advice from qualified counsel.

Primary sources

• FTC: Proposed policy statement addressing AI accuracy, July 1, 2026
• FTC: Operation AI Comply, September 25, 2024
• FTC: Final DoNotPay order, February 11, 2025