ASA filter rules for influencers are clear. You must not use a filter or editing effect that exaggerates the effect of the product you are promoting. Since 2021 the ASA has banned influencer ads for tanning products that used beauty filters, because the result no longer showed what the product could do. AI-generated images follow the same logic. There is no general UK duty to label AI use in ads, but an image must not mislead. CAP says a disclosure is very unlikely to fix an image that misrepresents results. France goes further and requires "retouched" and "virtual" image labels on influencer content. The EU AI Act adds deepfake transparency duties from August 2026.

This guide covers the ASA rulings, the CAP guidance on AI, the French labelling rules and how the AI Act overlaps, with a practical checklist for creators and affiliates. It is general information, not legal advice.

The ASA filter rulings and what they mean

In February 2021 the ASA ruled against two Instagram influencer posts for self-tanning products:

  • a post by Elly Norris for Skinny Tan, using a "Perfect Tan" filter; and
  • a post by Cinzia Baylis-Zullo for Tanologist, using a third-party beauty filter.

The ASA found both misleading because the filters exaggerated how effective the products were. The rulings followed the #filterdrop campaign started by make-up artist Sasha Pallari, and came with new CAP guidance on production techniques.

The principle is narrower than "no filters ever", but it is strict where it applies:

  • Directly relevant effects are the problem. A filter that smooths skin in an ad for foundation, whitens teeth in an ad for whitening strips, or deepens colour in an ad for tan exaggerates the product's effect.
  • Unrelated effects are lower risk. A colour grade on a background or a filter that does not touch the feature being sold is less likely to mislead. It still needs to be consistent with the claim.
  • Disclosing a filter does not cure it. If the image misrepresents the result, saying "filter used" does not make it accurate.

The same thinking applies to retouching in brand-made ads and to any product where appearance is the promise: cosmetics, haircare, fitness, wellness and some adult-wellness products. Our guide to #ad rules for influencers covers the disclosure side of influencer posts.

AI-generated ads: the CAP position

CAP published guidance on AI in advertising in 2025. Its main points:

  • No AI-specific disclosure rule. The CAP Code applies to an ad however it was made. There is no blanket requirement to say an image is AI-generated.
  • Two questions decide disclosure. First: is the audience likely to be misled if the use of AI is not disclosed? Second: if so, does a disclosure clarify the message, or contradict it?
  • Disclosure cannot rescue a misleading ad. CAP says disclosing AI is "very unlikely to mitigate the harm caused by a fundamentally misleading message". Its example is an AI-generated cosmetics image that does not reflect real results.
  • Disclosure can prevent a misleading impression where the ad is otherwise fair, for example by making clear that an influencer is AI-generated or that a deepfake is for comic effect.

In June 2026 the ASA reinforced this. It said AI-made ads face the same rules on misleadingness, offence and social responsibility as any other ad. AI depictions of real people must not imply a genuine endorsement. Advertisers stay responsible even when automated tools generate or place the ad. The ASA cited earlier rulings, including a 2023 ruling against Polyverse over a sexualised AI-generated image in an ad shown in a mobile game, and a 2025 ruling against Cosmos over an AI video app ad that generated intimate videos featuring a real singer. Both were found irresponsible and offensive.

For adult-industry marketers: AI-generated creatives that sexualise real people, or that appear in untargeted placements, combine every risk factor the ASA has flagged. Keep AI imagery away from real likenesses, and keep suggestive content on age-restricted inventory.

Virtual influencer rules

A virtual or AI influencer is a marketing communication when a brand pays and controls the content, just like a human one. The usual #ad rules apply. CAP's AI guidance adds that making clear the influencer is AI-generated can be necessary to stop readers assuming a real person used the product. If a virtual influencer "reviews" a skincare product, the claim that it "works" has no basis in experience and needs separate evidence.

Creators who use AI-generated personas on fan platforms face platform rules as well as advertising law. See our overview of AI content creator platforms.

Retouched images law in France and Norway

France's influencer law of 9 June 2023 requires influencers' commercial content to carry clear labels for the whole time the content is shown:

  • "Image retouchée" where image-processing has been used to slim or bulk out a silhouette or to change the appearance of a face; and
  • "Image virtuelle" where a face or silhouette has been produced by artificial intelligence.

The law applies to influencers whose audience includes French users, wherever they are based. Failing to label can lead to criminal penalties. France has separately required commercial photographs of models whose body shape has been edited to be marked "photographie retouchée" since 2017. Our guide to France's influencer law covers its other obligations.

Norway amended its marketing law in 2022 to require a label on paid social posts where a body's shape, size or skin has been retouched. If you run Nordic campaigns, check the current Norwegian Consumer Authority guidance.

How the AI Act deepfake duties overlap

Article 50 of the EU AI Act applies from 2 August 2026. It adds two transparency layers that sit alongside advertising law:

  • Providers of generative AI systems must mark output in a machine-readable way so it can be detected as AI-generated. Existing systems have until 2 December 2026 to comply.
  • Deployers who use AI to create deepfakes, meaning realistic images, audio or video of real people, places or events that could falsely appear authentic, must disclose that the content is artificially generated or manipulated. Lighter rules apply to evidently artistic, satirical or fictional work.

For a marketer, the AI Act disclosure is a floor, not a defence. An AI-generated testimonial video can carry an AI Act label and still breach advertising law because it implies a real customer experience. Read our full guide to AI Act deepfake labelling.

Checklist for filters and AI imagery in ads

SituationWhat to do
Beauty or appearance product, filter touches the promoted featureDo not use it. Show real, unfiltered results
Filter or grade unrelated to the product's effectGenerally acceptable. Check the overall impression
AI-generated "customer" or "before and after"Do not use it to show results. Use genuine, evidenced imagery
AI or virtual influencerLabel as an ad, make the AI nature clear, substantiate claims separately
AI likeness of a real personAvoid without written consent. Never imply an endorsement
Content seen by French audiencesAdd "Image retouchée" or "Image virtuelle" where the law requires
Realistic synthetic video or audio in the EUAdd AI Act deepfake disclosure, as well as meeting advertising rules

Next steps

Add a filter and AI question to your creative sign-off process and keep the unedited originals on file. For the rest of our marketing law coverage, browse the advertising rules section.