TikTok Shop UK bans fear-based organ imagery: lint your AI prompt

The UK Shop AI page bans fear-based health imagery like diseased organs. Check your prompt before rendering, then check frames with Sume video-frames.

5 min readSume
All posts

The UK TikTok Shop AI page lists fear-based health imagery, with diseased organs as the example, among prohibited AI content. A text-to-video prompt that asks for a glowing red liver or a clogged artery to sell a supplement is the kind of clip the page describes, so remove it before you spend on the render.

The page does not publish a list of banned words, so a prompt check is a habit, not a guarantee.

What the two Shop pages prohibit

The UK page also bans impersonating medical professionals, fabricating product capabilities, and altering size, packaging, features, and dimensions. The US page bars fake doctors and experts, exaggerated effects, false efficacy, and altering size, color, material, quantity, features, or compatibility.

Prompt patterns to rewrite (read 2026-10-07)
Prompt ideaPage concernSafer rewrite
Cutaway of a diseased organFear-based health imagery (UK)Show the product in use, calm light
Doctor in a white coat endorses itMedical-professional impersonationA plain narrator, disclosed as AI
Stain vanishes instantlyExaggerated effects, false efficacyShow what you actually tested
Mug twice its real sizeAltering size or dimensionsMatch the product page
Famous face holds the itemCelebrity impersonation (UK)An invented person

Check the frames after rendering

A prompt can pass and the frames can still surprise you. Video frames pulls up to 24 stills at times you choose, for sources up to 300 seconds, and always returns 202. Video inspect samples 8 stills by default, handles sources up to 1800 seconds, and can add a transcript so you can read what the voice claims.

Look at the stills for three things: a body-part close-up you did not ask for, text that contradicts the product page, and a product that has the wrong shape or color.

Fix and re-render

If a frame is off, rewrite the prompt and render again. Without a seed, you will get a different take, not a patched one. For a product with a wrong color in an otherwise good clip, an edit job may be cheaper than a new render. Keep the final job id and the stills in your ledger with the AI-generated toggle state.

A small prompt linter

You can automate a first pass. Keep a plain list of phrases that signal risk, such as organ names, "doctor says", "cures", "instantly", and brand or celebrity names, and flag any prompt that contains one. The list is yours, not TikTok's, so it will miss things, and a person should still read the flagged prompts.

Treat the list as a prompt review step before spending. A clip rendered from a flagged prompt costs money even if it never gets posted.

Look at the claim and the picture together. A calm shot of a supplement bottle with the line "supports daily wellness" is a different thing from a red, swollen organ with the line "fix this now". The UK page is concerned with fear used to push a sale, so a gentler approach is more than a rewording of the same idea.

Sources

Related posts

More in Use cases

All Use cases posts

Written by Sume