The UK Advertising Standards Authority has upheld complaints against three paid Meta ads for AI companion and image-to-video apps, told each advertiser the ad "must not appear again in the form investigated", and referred all three matters to CAP's Compliance team

Three rulings, one monitoring system, one publication date

The three rulings share an opening paragraph. Each Background section says the ruling "forms part of a wider piece of ongoing work on the advertising of AI products across a number of sectors" and that the ads "were identified for investigation following intelligence gathered by our Active Ad Monitoring system, which uses AI to proactively search for online ads that might break the rules". The ASA is describing its own tooling there, and we report it as the ASA's description.

All three are paid Meta ads. The tyan.ai ad, placed by an advertiser the ASA could only name as Scarlett, was seen on 10 March 2026 and promoted what the ruling calls "an AI companion generator app". The KH31 DD22 ad was seen on 2 April 2026 and the PictoPop ad on 31 March 2026; both promoted what the rulings call an "AI image-to-video generation app". The rulings were published together on 16 September 2026, which is the date each page prints under its heading.

These are adjudications, and that framing governs everything below. The ASA applied rules already in the CAP Code (Edition 12) to three identified ads. No ruling announces a new rule, a new policy for AI advertising, or a change to what the Code requires. The ASA is the UK's advertising self-regulator and the rulings do not discuss their own legal status; the sanctions they state are the ones in each Action section, set out later in this piece, and we attribute nothing beyond those.

Each page also links related rulings published the same day, including two against advertisers named as Guangxi Baise Tianyi Agricultural Development Co Ltd and Rusto AI-AI Photo Toolbox. We did not review those two and say nothing about them.

What the ASA found in the tyan.ai ad

Only the Animcha ruling carries two investigated issues, and its Summary of council decision says both were Upheld. The Ad description records a video in which a young woman in a red bikini, hair in pigtails tied with pink bows, sits in a room containing "shelves, books, teddy bears and other plush toys" and says "Have you ever dreamed of having your own personal AI assistant, ready to follow your orders?". The voice-over that follows includes "I am your AI girl, designed to follow your instructions" and the ad returns to its opening scene with "Create your own AI girlfriend now."

On the first issue, the ASA considered the ad portrayed the woman as "a submissive and compliant sexual object" and reinforced the harmful impression that it was acceptable to objectify and stereotype women. It concluded the ad breached rules 1.3 (Social responsibility) and 4.1 and 4.9 (Harm and offence).

The second issue is the one specific to this ruling. The Assessment records that the woman's "youthful facial features", the pigtails with bows, the setting that "gave the appearance of a child's bedroom", a Hello Kitty top, and nightwear reading "Daddy's girl" worn in "a child's bed with a night light and teddy bear" were, in the ASA's view, "likely to lead many viewers to interpret her as being under 18 years of age". Because the ad depicted a person who appeared to be under 18 in a sexualised way, the ASA concluded it was irresponsible and breached rules 1.3, 4.1 and 4.8. The finding is about apparent age. The ruling makes no finding about the actual age of any person, and we make none either.

The Assessment also records that Animcha Ltd t/a tyan.ai and Person(s) unknown t/a Scarlett did not respond, and that Scarlett failed to provide a full name and geographical business address. The ASA treated that as breaches of rules 1.7 (Unreasonable delay) and 1.7.1 (Compliance) and told them to respond promptly in future. The respondent panel gives Animcha Ltd an address in Nicosia, Cyprus.

The image-to-video ads: objectification plus a consent finding

KH31 DD22 and PictoPop were each investigated on a single issue, whether the ad sexualised and objectified women and was therefore irresponsible, offensive and harmful. Both were upheld on rules 1.3, 4.1 and 4.9. What makes these two rulings worth reading closely is the second step in the ASA's reasoning.

In the KH31 DD22 ad, a woman in a white cropped top and pink skirt lifts her top and exposes her breasts; the ad then shows a phone interface, a "Before" image of the same woman clothed, a "createVideo" button, and an "After" version in which she is exposed, under the caption "People are quietly using AI to transform photos". The ASA considered viewers would understand the app "was capable of generating videos from an image, including by removing clothing and creating sexually explicit content at the user's discretion". It then went further: because the ad implied users could select any photo from which a sexually explicit video could be created, "we considered the ad condoned the digital manipulation of women's images to create sexually explicit content without their consent".

The PictoPop ruling reaches the same conclusion in almost the same words. The ad showed a woman kneeling in front of a standing man and unbuttoning his trousers, with the overlay "turn photos into hot motion" and the caption "PictoPop lets adults explore bold, creative photo transformations in seconds. Easy. Addictive". The ASA considered the phrase "Easy. Addictive" reinforced the impression that the app could be used easily to create such content for the user's sexual gratification, and that the ad "condoned the digital manipulation of women's images to create sexually explicit content without their consent".

Two limits on that finding are worth stating. First, it is a finding about what the ads condoned, drawn from the ad content. Neither ruling tests the app, verifies that it produces such output, or identifies any real person whose image was manipulated, and we claim none of those things. Second, the PictoPop ruling records an App Store claim of "1M+ downloads" as part of the ad description only; the ASA neither verified nor relied on that figure, and neither do we.

On the procedural side, KH31 DD22 did not respond and did not supply a name or address, which the ASA recorded as breaches of 1.7 and 1.7.1; the ASA could name the advertiser only as Person(s) unknown. Sinosea Technology PTE Ltd t/a PictoPop did not respond either, recorded as a breach of 1.7. Its panel gives an address in Singapore. That is a mailing address printed by the ASA, and we do not treat it as evidence of where the company is registered.

Nobody answered, and Meta says it had already acted

Short and nearly identical, the three Response sections tell one story. In each, the advertiser did not respond to the ASA's enquiries. In each, Meta confirmed the ad had appeared on its platform, said it breached Meta's advertising guidelines and standards, and said the ad "had already been removed before they were contacted by the ASA".

That chronology is Meta's statement as recorded by the ASA, and the ASA makes no finding on it. The sequence the rulings give is removal first, ASA contact second, ruling on 16 September. Nothing in the rulings says the ASA's investigation caused the removals, and nothing says the ads or copies of them are gone from anywhere else. Meta is not the advertiser under investigation in any of the three rulings, no finding is made against it, and no direction in any Action section is addressed to it.

The panels are also consistent in one small detail: each records one complaint and lists the agency as none, unconfirmed. Each Issue section opens with "The ASA challenged whether", which is the ASA's own framing of the matters it investigated.

What the Action sections do, and what they do not

Every one of the three Action sections has the same three parts. The ad "must not appear again in the form investigated". The advertiser is told to ensure future ads are socially responsible and do not cause serious or widespread offence, "including by featuring a harmful gender stereotype that objectified and sexualised women". And the matter is referred to CAP's Compliance team. The Animcha ruling adds one more sentence, telling the advertisers not to represent anyone who was, or seemed to be, under 18 in a sexual way.

All three rulings contain the referral in the same words. The rulings do not say what the Compliance team will do, and we do not say either. No ruling imposes a fine, prohibits the app from operating, restricts the underlying product, or sanctions Meta. A reader who wants those outcomes will not find them in these documents.

What the rulings do establish is narrower and still useful. Three specific ads for AI apps, placed as paid Meta ads reaching UK users, were tested against CAP Code rules that already existed, and failed. An advertiser drafting a demo-format ad for a photo-to-video tool now has two written examples of the reasoning the ASA applies to that format, and a companion-app advertiser has one written example of how the ASA reads a childlike setting combined with sexualised presentation.

What we did not verify

Typography: where a quoted ASA passage contains a typographic apostrophe in the source, it is rendered here as a straight apostrophe; no other character is changed. What we opened: the ASA's published ruling on Animcha Ltd t/a tyan.ai at asa.org.uk, read end to end from the Background through the Action section and the respondent panel carrying Complaint Ref A26-1341105; and the ASA's published rulings on Persons unknown t/a KH31 DD22 (Complaint Ref A26-1342725) and Sinosea Technology PTE Ltd t/a PictoPop (Complaint Ref A26-1342721), read the same way from retained copies of the ruling pages. Every quotation above was checked character for character against the page text.

What we did not open: the CAP Code (Edition 12) itself, so the content of rules 1.3, 1.7, 1.7.1, 4.1, 4.8 and 4.9 is stated here only as each ruling paraphrases it; the ASA rulings index for 16 September 2026; the related rulings against Guangxi Baise Tianyi Agricultural Development Co Ltd and Rusto AI-AI Photo Toolbox; the ads themselves, the apps, Meta's advertising standards, and any company register entry for Animcha Ltd or Sinosea Technology PTE Ltd.

What we refuse to claim: we do not say these rulings create any new rule or extend the CAP Code to AI advertising, because they apply existing rules to three ads. We do not say any app was banned, any fine was imposed, or any sanction was applied to Meta, because no ruling says so. We do not say any real person under 18 appeared in the tyan.ai ad; the finding is that the character seemed to be under 18. We do not say the apps actually generate non-consensual sexual content or that any person was harmed by them; the finding is that the ads condoned it. We do not say the ASA caused Meta's removals, or that the ads have disappeared everywhere. We do not treat the Singapore address as proof of registration, we do not repeat the "1M+ downloads" claim as a fact, and we give no total for the number of AI rulings the ASA published that day. Where the word must appears above, it is quoting the rulings.

Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel.

Related briefings

• UK ASA makes AI a 2026 priority: disclose AI in ads where hiding it would mislead
• Weekly enforcement roundup
• AI disclosure rules tracker
• Europe AI regulation tracker
• AI Regulation News hub