TikTok AI Content Checker and Likeness Detection, Explained
TikTok's free AI checker reads its own watermarks on uploads, even without an account. Likeness Detection starts a global, opt-in rollout.

TikTok's free AI checker reads its own watermarks on uploads, even without an account. Likeness Detection starts a global, opt-in rollout.

On November 20, 2026, TikTok's newsroom published More tools to provide clarity and control around AI-generated content ↗. There are two tools. The first is a public checker that tells you whether an image or video was made or edited with TikTok's AI creation features. The second is Likeness Detection, an opt-in tool for eligible creators, which TikTok is "starting to globally roll out." Euronews ↗ (Jonathan Benton) and RTÉ ↗ covered the launch the same day. The lead image is TikTok's own newsroom artwork for the post, used for news reporting; credit Image: TikTok. We have not uploaded anything to the checker.
The checker lives on TikTok's Transparency Center page Identifying content made with TikTok AI ↗. You do not need a TikTok account. You upload one file at a time, and the tool looks for two signals that TikTok adds to content made with its AI features:
C2PA Content Credentials, metadata that other platforms can read too. The checker only reads credentials added by TikTok.
VeraDura, TikTok's own invisible watermark, which TikTok says is "more durable and harder to remove."
If a signal is intact, the page says you will also see which TikTok tool was used. Euronews reports that the checker covers ByteDance's CapCut, Dreamina, Hypic, and Pippit apps as well as TikTok.
| Media | Formats | Max size |
|---|---|---|
| Images | JPEG, PNG, WebP | 20 MB |
| Videos | MP4, MOV, WebM, AVI | 100 MB |
The page's intro also mentions audio, though its format list shows only images and videos. Euronews puts audio files at the same 100 MB cap as video.
Read a "no signal" result carefully. TikTok's page says it can mean the file was not AI-made, or was made with another company's AI tools, or that the signal is weak: credentials stripped, watermark degraded, or a file older than these signals. TikTok's own words: "no AI detection tool is 100% foolproof." For credentials added by other platforms, TikTok points to the Content Authenticity Initiative's verify tool ↗.
On privacy, TikTok says it deletes your upload "immediately after showing you the result," and never uses it to train models.
To use it: open the Transparency Center page, upload the image or video, and read the result. You can leave anonymous feedback on the results page.
TikTok lists three steps. Eligible creators verify their ID with a "trusted third-party partner." TikTok then notifies them when content and accounts on TikTok may be using their likeness, which they review inside the tool. They can report that content to TikTok's safety teams, and TikTok removes content that breaks its rules on harmful, misleading, or nonconsensual depictions.
The tool is for "eligible creators and public figures," including government, politician, and political party accounts. TikTok does not spell out the eligibility bar in the post. It says Likeness Detection is designed to minimise data collection, and that TikTok does not keep ID documents. People who do not qualify, and people who are not on TikTok at all, can still report nonconsensual AI content through TikTok's privacy reporting form.
RTÉ frames the launch around elections: it quotes Francesca Scapolo, TikTok's government relations lead for European elections, saying "Protecting the integrity of the elections is one of the most critical responsibilities that we have at TikTok." Euronews notes that EU AI Act transparency rules, including labels for deepfakes, began rolling out on August 2. RTÉ adds that YouTube has built its own likeness detection tool for creators.
TikTok's post also repeats some of its own numbers: over 5 billion AI-generated videos labelled to date, more than 99% of rule-breaking content removed before anyone reports it, and more than 4 million people helped by its AI literacy hub. Those are TikTok's figures, not independently checked.
"A global likeness rollout is starting for eligible creators, and it is opt-in rather than a full launch." — The Circuitry ↗ (@thecircuitry_)
"A useful step, but enterprises face a much broader problem." — Orivfy ↗ (@Orivfy_)
"Proof carried inside real clips would shorten the hunt" — M Thomas George ↗ (@mthomasgeorge)
Reddit was not reachable for this story.
The checker answers one question well: did a TikTok or ByteDance tool make this? That is useful for a big slice of viral video, since CapCut edits travel far beyond TikTok. It is not a general deepfake detector. A clip from another company's generator will come back with no signal, and so will a TikTok clip whose metadata was stripped. Treat "no signal" as "unknown," never as "real."
Use it next to Content Credentials, not instead of them. For a suspicious file, run TikTok's checker and the CAI verify tool. Two narrow checks beat one.
Likeness Detection matters most to public figures for now. Politicians and large accounts get alerts and a review queue. Everyone else still has the report form. Watch for the eligibility rules and how fast reports are handled; the post gives neither.
If you make AI video yourself, expect labels and watermarks to follow your exports. Our AI video generators guide and free AI photo editor review cover the tools, and this week's Kling Image 4.0 note and Canva AI off switch note cover two other AI image moves.
Verdict: worth bookmarking if you check viral clips for a living, and a good opt-in for public figures. It will not settle whether a video is real.