In December 2025, FTC staff sent warning letters to ten companies about potential violations of the agency's new Consumer Review Rule, which bans fake and false reviews, including reviews generated by AI. Warning letters are not enforcement actions. But they are a clear signal: the FTC is watching how businesses use AI-generated content in marketing, and it is willing to remind companies, in writing, that violations can carry civil penalties of up to $53,088 per violation.
Here's the actual legal framework, how it applies to common fashion brand scenarios, and what disclosure language satisfies the standard. One honest note up front: the FTC has not issued a rule that says every AI-generated ad image needs a label. What it has is Section 5, and Section 5 is enough to reach most of the conduct that should worry you.
The legal framework: deception, not technology
Three instruments matter.
First, Section 5 of the FTC Act prohibits deceptive acts and practices in commerce. A representation is deceptive if it is likely to mislead reasonable consumers about something material, something that would affect their decisions. This standard is technology-neutral. It applied to airbrushing. It applies to AI.
Second, the FTC's Endorsement Guides, 16 CFR Part 255, updated in June 2023, require disclosure of material connections between advertisers and endorsers. The 2023 revision matters for AI specifically because the updated definition of "endorser" covers fabricated and non-existent endorsers, including virtual influencers. A synthetic persona promoting your brand is inside the Guides, not outside them.
Third, the Consumer Review Rule, 16 CFR Part 465, effective October 21, 2024, prohibits fake reviews and testimonials, explicitly including AI-generated ones, and lets the FTC seek civil penalties. This is the rule behind the December 2025 warning letters.
So when does AI use become a disclosure problem? When the AI content misleads consumers about something material: what the product looks like, whether the person recommending it exists, whether a review reflects a real experience. If consumers would act differently knowing the truth, the truth must be disclosed. That is the FTC's long-established standard applied to a new context.
The scenarios that matter in fashion
AI-generated product imagery that doesn't match the product. A garment depicted on Instagram looks real because AI rendered it photorealistically, but the garment was never photographed, and the colours, textures, and drape shown don't match what will ship. You don't need an AI-specific rule to see the problem. This is a misrepresentation of the product itself, classic Section 5 territory, and the safest cure is both fixing the imagery and disclosing the render.
Synthetic influencers presented as real people. AI-generated social media personas with realistic profile photos, backstories, and posting histories promoting fashion brands, while the audience believes they're following a real person's style journey. Under the 2023 Endorsement Guides, a fake endorser is a deceptive endorser. The material connection to the brand must be disclosed, and so must the fact that the persona isn't a person.
AI-generated reviews. Customer reviews generated by AI and presented as authentic feedback. The Consumer Review Rule makes this a penalty offense regardless of whether the reviews happen to be accurate. In 2024's Operation AI Comply sweep, the FTC went after Rytr, an AI writing tool whose review generator produced detailed testimonials untethered from any real experience. Worth knowing: the Commission set aside the Rytr order in December 2025 as part of the administration's AI Action Plan, a signal that it will target businesses that deploy fake reviews rather than the tools that could make them. The Consumer Review Rule itself is unchanged.
Photorealistic synthetic models in campaigns. This is the greyest federal zone. If the AI model isn't misrepresenting the garment and isn't posing as a specific endorser, current federal law does not clearly require an "AI-generated" label. But the states are filling that gap: New York's synthetic performer disclosure law took effect June 9, 2026 and requires conspicuous disclosure when ads reaching New York audiences feature AI-generated synthetic performers, with penalties of $1,000 for a first violation and $5,000 for each subsequent one. I break that law down in my New York explainer, and the EU AI Act imposes its own disclosure duty from August 2026. Disclosure is rapidly becoming the default across jurisdictions. I track the state-by-state picture on the tracker.
What disclosure actually looks like, and what doesn't work
The FTC's standard for adequate disclosure: clear, conspicuous, and unavoidable at the moment the consumer encounters the content. The 2023 Guides are explicit that a disclosure buried behind a link, in a bio, or in terms of service doesn't count. It must be understandable to ordinary consumers, not just people fluent in AI terminology.
What works:
In social media posts: "AI-generated image" or "Created with AI" in the post caption before any "see more" truncation, as a text label on the image itself, or both. Hashtags are riskier. The Guides treat ambiguous or buried hashtags skeptically, and my advice is plain text over #AIgenerated every time.
In advertising creative: a visible text label on the ad unit, not exclusively in fine print, identifying AI-generated elements.
On e-commerce product pages: where AI-generated imagery is used instead of photography, disclosure on the product page proximate to the imagery. "AI-generated visualisation, actual product may differ" is honest and clear. Our disclosure generator will draft channel-specific language for you.
What doesn't work: disclosure buried in website terms of service or privacy policies. Disclosure made only in brand FAQs or "about" pages rather than proximate to the specific content. Disclosure that only appears on hover or in collapsed content most consumers won't encounter.
The test is not whether a disclosure exists somewhere in your content ecosystem. The test is whether a reasonable consumer encountering this specific piece of content would see and understand it.
Synthetic influencers: a specific compliance framework
Brands using AI-generated personas have a specific challenge: the persona's AI nature must be disclosed in a way that reaches the audience, not just in meta-level brand communications that most followers will never read.
Compliant synthetic influencer practice requires:
Clear identification in the account bio that the account represents an AI-generated persona. Not buried in the third paragraph. A prominent bio statement.
Disclosure in each piece of sponsored or branded content that the persona is AI-generated. The brand partnership disclosure and the AI identity disclosure are separate obligations and both apply.
No presentation of the synthetic persona as a real person, including in press materials and partner communications, anywhere the audience might reasonably believe they're engaging with a real individual.
The FTC's leadership and priorities shift. Its deception authority doesn't. Build your disclosure practice around the consumer's understanding, not around the enforcement mood of any given year, and you'll be positioned for the federal standard, the state laws, and the EU all at once.
This article is editorial analysis, not legal advice. For questions about your specific situation, consult a qualified attorney.