FTC AI Disclosure Rules 2026: Startup Fix Guide

FTC AI Disclosure Rules 2026: What Changed and How to Fix It Fast

Short answer: the FTC already treats hiding where content came from as deceptive when that origin would change how a reasonable person judges it, and its Consumer Review Rule bans fake reviews and testimonials, including ones written by AI. If your startup uses AI for landing page copy, review summaries, or affiliate content and you have not added specific disclosures, you are exposed. Fixing it takes a sprint, not a quarter.

I work with founders building for the US market from outside it, and I want to be upfront: I am a growth operator, not an attorney. This is an operational checklist built from what founders are asking me on calls this month, not legal advice. Talk to counsel about your specific jurisdiction before you finalize anything here.

What Actually Changed, and What Has Not?

The FTC’s Consumer Review Rule (16 CFR Part 465) has been in force since October 2024. It bans buying, selling or creating fake reviews and testimonials, AI-generated ones included, and the FTC has since begun sending warning letters under it. In 2026 the agency also asked for public comment on how companies make claims about AI; that is a proposal, not a rule in force. Nothing here bans AI. What changed is how seriously AI-made reviews and testimonials are treated. Check the FTC’s own site for the current text before you rely on any of it.

Most startups have treated AI disclosure as a nice-to-have footer line. The FTC’s long-standing position is that hiding the origin of content, when that origin would change how a reasonable person judges the content, is deceptive on its own. A true product claim written by AI but dressed up as a “customer’s honest review” is now a problem even if every fact in it checks out.

If you are a founder outside the US selling into the US, this matters more than it might seem. You are already fighting an uphill battle to earn trust with American buyers who cannot meet you in person. An FTC warning letter, even one that never becomes a fine, is the kind of thing a US enterprise buyer’s legal team will find in five minutes of due diligence.

Are My AI-Written Landing Pages a Ticking Time Bomb?

Not automatically. But if your landing page copy makes efficacy, price, or comparison claims, two things now have to be true at once. First, you need to be able to substantiate the claim the same way you would if a human wrote it. Second, the AI use itself cannot mislead someone into thinking a real person is speaking from personal experience.

This is where I see the most exposure with consumer AI founders. You ran your copy through ChatGPT or Jasper to speed up a landing page for a US audience, and the model, doing what it does well, wrote a warm first-person case study voice: “When we switched to this tool, our support tickets dropped.” Nobody switched to anything. That sentence was generated. If a reader would reasonably believe a real customer said that, you have a problem regardless of whether the underlying claim is directionally true.

The fix is not to stop using AI for speed. It is to audit for a specific pattern: any block of text written in first person, past tense, “personal story” voice that was not actually lived by a real person. Flag it, then either replace it with a real testimonial, rewrite it in a way that does not imply lived experience, or disclose that it is AI-assisted marketing copy rather than a customer account.

How Do I Disclose AI-Generated Reviews and Testimonials?

A generic line buried in your footer or terms page saying “we may use AI on this site” does not meet the bar. The FTC’s expectation is “clear and conspicuous,” meaning the disclosure sits next to the thing it is disclosing, not three clicks away. If your product page shows a 4.6 star AI-summarized rating, the disclosure needs to sit right next to that number, something like “This summary was generated by AI from customer reviews; individual reviews may vary.”

The risk I would look at first is the AI-generated star rating itself. If your system takes raw review text and generates a synthesized score or summary without disclosing that a model did the synthesizing, a buyer can read a model’s summary as real customers’ verdict, which is exactly the impression the FTC’s rules on reviews are about, even when the underlying reviews are real. This hits a lot of consumer AI founders who use review aggregation tools to build trust signals on pricing pages, which is exactly where a US buyer is deciding whether to trust an unfamiliar international vendor.

On mobile specifically, “clear and conspicuous” generally means visible without the reader tapping “more” or scrolling past the fold to find it. If your disclosure only shows up after someone expands a review, you have not met the bar on the device most of your US traffic is using.

What Do My Affiliates and Influencers Need to Disclose About AI?

This is the part founders miss most, and it is the one I’d flag first if you are running any affiliate or influencer and UGC motion through networks like Impact or ShareASale. You are liable for what your affiliates do wearing your brand’s name, even when you did not write the content yourself.

If an influencer uses an AI-generated script to promote your product, two disclosures need to happen, not one. They need to disclose the affiliate or paid relationship, which most influencers already know from existing FTC endorsement rules. And separately, if the content implies a human did hands-on analysis or testing that an AI script actually produced, that needs disclosure too. “Paid partnership” does not cover “AI wrote this and I read it on camera.”

For founders running remote, distributed teams managing US affiliate programs across time zones, this is a contract problem before it is a content problem. Update your affiliate agreement now with a specific clause requiring AI-use disclosure in any post promoting your product. Do not wait for your affiliate manager to catch it during a content review, because by the time it is published, the exposure already exists.

**Want a live breakdown of your startup's growth?**

I help early-stage consumer AI founders escape the “great product, zero distribution” trap.

Work with me →

How Do I Keep Growth Experiments Running Without Breaking These Rules?

You do not need to slow down testing. I would actually push back on any founder who tells me they are pausing AI-driven growth experiments entirely over this. That is an overcorrection that costs you a growth quarter for no real safety gain.

What you need is one more variable in your experiment tracking. Tag every test with two booleans: requires_substantiation and requires_ai_disclosure. If either is true, the disclosure gets built into the variant before it ships, not bolted on after legal flags it. This is the same discipline you already apply to load time or mobile rendering checks before a variant goes live, just applied to a new risk category.

One thing worth measuring honestly: adding a disclosure line next to a star rating or review summary can move conversion, sometimes down slightly in the short term. I would rather see a founder A/B test the exact wording and placement of a compliant disclosure than skip it and hope no one notices. Treat the disclosure copy itself as a testable variant, because in most cases a well-placed, honest disclosure costs you less conversion than founders fear, and it costs you a lot less than a warning letter that a US enterprise prospect’s legal team finds during due diligence.

The 5-Step Compliance Quick Check for Your Next Sprint

Here is the version you can actually run in your next sprint, in order of how fast each one takes.

1. Audit all AI-written product copy for “personal experience” language. Anything that reads like a lived testimonial but was not, gets flagged and rewritten or disclosed.
2. Add an inline disclosure to any AI-generated review summary or star rating module. It sits next to the number, not in a footer.
3. Update affiliate and influencer contracts to require disclosure of both the paid relationship and any AI-generated script or content.
4. Review your UGC and case study sections for content that was AI-generated rather than genuinely user-submitted, and label it accordingly.
5. Update your privacy policy to state plainly that AI is used in content generation, review summarization, or copywriting on the site.

None of this requires a legal team you do not have yet. It requires an afternoon with your growth lead and a shared doc.

What This Means If You’re Selling Into the US From Outside It

I’ll say the part founders usually ask me on the call but hesitate to type in an email. If you are running a consumer AI company out of India or Europe with a remote team, you are already worried whether a US buyer trusts your product’s context. An FTC compliance gap does not just create legal risk, it creates a trust gap at the exact moment a US enterprise buyer is deciding whether an international vendor understands their market. Clean, specific disclosures are one of the few things you fully control here. Get that right and it is one less reason for a skeptical buyer to walk.

If you are early stage and pre-revenue, do this yourself. It is a checklist, not a consulting engagement, and paying anyone to run this audit for you right now would be premature. If you are post-seed to Series A and running paid growth experiments across a US affiliate network with a remote team spread across time zones, that is usually when it is worth a working session to build the substantiation and disclosure checks directly into your experimentation process rather than retrofitting them later.

If you want to work through your specific setup, book a call with me and bring your current landing pages and affiliate agreements. We will go through what needs to change before your next sprint, not after a warning letter does it for you.