If you sell a US TikTok Shop brand or mature team running affiliate creators at volume, with legal review in the process and more than a hundred videos shipping a month, the constraint that decided this year was not how much content you could produce. It was whether what the content says can be proved. The revised FTC Endorsement Guides took effect on 1 March 2026, and the three changes that matter for a creator programme are specific rather than general. First, paid amplification creates the disclosure obligation: putting spend behind a creator's post through tools such as TikTok Spark Ads is a material connection that has to be disclosed whether or not the creator was paid, and the platform does not add the label for you when it amplifies an organic post — that is the operator's job. Second, review gating is now explicitly deceptive: steering satisfied customers to public reviews while routing unhappy ones to private feedback is a practice the guides name, not a grey area. Third, AI-generated endorsement material has to be labelled inside the content itself — synthetic reviews, AI voiceover video and AI-written quotations cannot sit in a terms-of-service line. Alongside those, the placement standard has teeth: a disclosure buried at the end of a caption, inside a wall of hashtags after the more cutoff, or in a text overlay that appears for under three seconds does not meet clear and conspicuous. And the exposure is no longer only the creator's — if an affiliate creator fails to disclose and the brand knew or should have known, the brand carries the risk directly.
Price it in the units you actually lose. Take a US brand doing 2,000 orders a month at $32 average order value, so $64,000 of GMV, with affiliate-driven sales at roughly 42% — about $26,880 coming from creators across 120 videos a month. If 15% of those videos ship with a missing disclosure or an unlabelled AI element, that is 18 assets. Three costs stack. The videos get quietly removed, which at an average $224 of attributed GMV each is about $4,032 gone. The creators attached to them get flagged, which is a retention cost rather than a one-off: an account that has been marked once stops taking your briefs. And the ceiling is not theoretical — in the second quarter of 2026 alone the FTC issued $14.2m in civil penalties across five DTC brands, which averages $2.84m each, and a single penalty of that order against a $64,000-a-month business is not a cost of doing business. Running the check below takes about ninety seconds per video: three hours a month against a hundred and twenty assets.
What changed on 1 March 2026, and why it reaches the brand
The scale is the reason this cannot be left to the creator's judgement. US social commerce GMV passed $120bn in the first half of 2026, and the enforcement attention follows the money: in the second quarter of 2026 the FTC issued $14.2m in civil penalties across five DTC brands. What makes it reach the brand rather than stopping at the creator is the revised guidance on material connections and on brand responsibility — if an affiliate creator fails to disclose and the brand knew or should have known, the brand is exposed directly rather than vicariously. The operational detail that trips most teams is Spark Ads. Amplifying a creator's organic post with paid spend creates the connection, and the platform does not add a disclosure when it does so, so a post that was compliant when it was published becomes non-compliant the moment someone puts budget behind it. That is a change made by your media team, not by your creator, and it is invisible in the asset itself.
Two platform rules sit underneath the legal one and are easier to miss. TikTok requires a spoken disclosure in the video, not only a caption hashtag, with the verbal line in the first three seconds and #ad or #sponsored in the caption. Separately, the platform's AIGC rule requires an AI-generated content label when AI produced the script or the visuals, and TikTok Shop's high-quality LIVE requirements prohibit AI voiceover or pre-recorded audio in promotional livestreams — that last one is enforcement, not interpretation. None of these substitute for the FTC standard, and none of them are satisfied by it. You have to meet both.
Step 1 — Turn the requirement into seven checks on one page
Write the requirement as seven lines on one page, in the order a video is made, because a policy document is not a check anyone runs at 6pm. One: the platform's branded content toggle is on, not just a hashtag in the caption. Two: the spoken disclosure lands in the first three seconds and before the product link or purchase prompt. Three: the caption's first line carries #ad or "Paid partnership with [brand]" ahead of the more cutoff. Four: if AI touched the asset, the platform AIGC label is on and the caption states what AI did. Five: no synthetic presenter is implying a personal experience it cannot have had. Six: no AI voiceover or pre-recorded audio in a promotional livestream. Seven: the contract behind the creator has disclosure, AI-declaration and clawback clauses. One page, one owner, one pass per asset.
The wording does not need to be clever, and clever wording is how disclosures fail the standard. The FTC's own plain-language examples are the model: "This is an ad for BRAND" and "BRAND paid me to tell you about it." For AI, name the specific use rather than the category — "This video uses an AI voiceover" or "The presenter in this video is AI-generated" or "AI-assisted editing" tells a viewer something actionable, while a general note that AI tools were used somewhere in the workflow does not.
Seven lines on one page is the whole artefact, and it only works if it sits where the asset is approved rather than where policy is stored. register with DAMI and ask the team for the US pre-publish disclosure checklist so each SKU carries its own version and the check has a place to be recorded instead of a person to remember it.
Step 2 — Keep the paid disclosure and the AI label as two separate obligations
The most common error is treating one label as covering both obligations. An AI-generated label does not tell a viewer you were paid, and a paid-partnership toggle does not tell a viewer the presenter is synthetic, so a sponsored video with an AI presenter needs both. Order them rather than stacking them: lead with the paid disclosure, because that is the one that answers the buying question, then apply the platform's native AI label, which is designed to be visually distinct. Where the format is voiceover-led, say both out loud, because a viewer who never reads the caption still hears the line. Do not rely on a profile-level note that the creator sometimes uses AI tools — that does not discharge a per-post obligation under either standard.
Watch the middle of the calendar, not the ends. Nobody fails to label a fully synthetic spokesperson — the trouble is AI-assisted work that looks like a minor edit: upscaled footage, an AI voiceover overdubbing a language version, a generated background, or an auto-summary that drops the #ad line that existed in the original. If AI meaningfully altered what viewers see or hear, the label applies. Build the check around that question rather than around whether a tool was formally used.
Step 3 — Put the obligation in the contract, not only in the brief
A brief is not enforcement, because a brief only reaches the creator who reads it. Three clauses turn the obligation into something you can act on: a representation that the creator will comply with FTC disclosure requirements; a declaration, before delivery, of any AI-generated or AI-assisted element, with a named party responsible for applying the platform's AI label; and a clawback provision so that penalties arising from the creator's non-compliance can be recovered from commission. The point of the second clause is timing — finding out that AI was used after the video is live is the most expensive moment to find out, because the fix is a takedown rather than a label.
Assign the label responsibility explicitly rather than assuming it. If the brand's social team publishes and the creator shoots, the party who applies the platform AI label is whoever holds the account at publish time — and that is often not the person who knows whether AI was used. Write the name into the clause. An unassigned obligation is an obligation that is met in month one and skipped in month four.
Step 4 — Sample what is already live, because gates decay
Sample what is already live rather than trusting the gate. Once a week, pull ten live creator videos and run the seven checks as if you had never seen them, recording pass or fail per line. What you are measuring is not whether the creators understood — it is whether the check is still being run, and specifically whether the Spark Ads case is covered: a video that passed at publication can fail the moment media spend is attached to it. Track the failure rate by line, because the pattern tells you which check is being skipped, and a single line failing repeatedly is a training issue rather than a compliance one.
Connect the check to the score that actually gates you. Shop Performance Score runs 0 to 5 across six inputs calculated on the last 60 days, including after-sales handling time, and it determines access to affiliate marketing, campaign participation and accelerated settlement. Since 25 February 2026 US sellers must use TikTok Shop Logistics Services or approved providers, so fulfilment is no longer a variable you control freely either. Disclosure failures and fulfilment delays now feed the same score, which is why the seven checks are worth running as one pass rather than as two separate reviews in two teams.
How DAMI makes the check a field instead of a reminder
DAMI brings an 8-million-plus creator database, competitor and competitor-store creator search, automated outreach tasks, AI-assisted outreach, AI video creation, sample management, campaign management, social media management, store analytics, and multi-store team dashboards into one place. Three actions put the check where the asset is approved, and each one ends in something you can audit rather than something you have to trust:
Make the check a status field on the asset, not a message
Manage creator content in DAMI's social media management with the seven checks as a status on each asset, carrying who approved it and when. The output is a per-asset record that shows the review happened, which is the only thing you can produce when a video is questioned later.
Ship the compliant reference version so the creator starts from it
Use DAMI's AI video creation to produce a reference cut per SKU with the spoken disclosure in the first three seconds and the caption line already in place, then hand it over as a starting point. The output is a compliant reference asset per SKU, and the disclosure stops depending on the creator remembering.
Review disclosure and takedowns by creator and by SKU every month
Use DAMI's store analytics to break video takedowns and disclosure failures down by creator and by SKU, then rank them. The output is a monthly table that tells you which creator relationships and which product lines are carrying the compliance cost, and where to spend the training.
Frequently asked questions
If the creator fails to disclose, is it their problem or mine?
Plan as though it is yours. Under the revised guidance the brand carries direct exposure where it knew or should have known that an affiliate creator did not disclose, so the exposure is not merely vicarious. That is also why the contract clauses matter — a clawback provision at least converts the cost from unrecoverable to recoverable.
Does the AI label satisfy the disclosure requirement?
No, and this is the substitution error that causes most failures. An AI-generated label tells a viewer the presenter is synthetic; it does not tell them you paid. A paid-partnership toggle does the reverse. A sponsored video with an AI presenter needs both, applied separately, with the paid disclosure leading.
Is a small AI-assisted edit enough to require the label?
If AI meaningfully altered what viewers see or hear, yes. Upscaled footage, an AI voiceover for a language version, a generated background, or an auto-summary that strips a disclosure line all fall inside that test. The risk is concentrated in the middle of the calendar, not in the obvious fully synthetic cases.
Can I run the check once and rely on it?
No, because the asset changes after approval. Attaching paid spend to an organic post through Spark Ads creates a material connection that has to be disclosed, and the platform does not add the label when it amplifies. A video that passed at publication needs the check again the moment media spend is attached.
Can I manage the seven checks in a shared sheet?
For the first month, yes. After that the sheet stops being evidence: it does not show which version of the check applied, it cannot tell you that a video changed status after approval, and it records a pass once rather than per state. The failure is not that the sheet is wrong — it is that in the week volume peaks, the approval is the step that gets skipped, and the sheet still shows last week's passes.
Conclusion
Authenticity is the one input that decides whether your content is allowed to keep running, and everything above is about making every video hold up on the day somebody checks it. DAMI brings an 8-million-plus creator database, competitor and competitor-store creator search, automated outreach tasks, AI-assisted outreach, AI video creation, sample management, campaign management, social media management, store analytics, and multi-store team dashboards into one place.
Pull ten live creator videos and run the seven checks against them as if you had never seen them, recording pass or fail per line. Fifteen minutes, and the line that fails most often is the one to fix first — not the one you assumed was the problem.
Start with DAMI and put the check where the content is approved rather than where the policy is stored.
