Rhinehart Studio

Home › Guides

Guide · UK and US

Using AI-generated images and video in your Christmas ads: the UK and US rules for 2026

Generative AI makes it quick to put your product in a snowy shop window or have a friendly presenter talk through your gift guide. Neither the UK nor the US bans that. What both ban is an ad that misleads, and AI makes some old mistakes very easy to repeat. Here is what matters for AI images and video in your Black Friday and Christmas 2026 ads, each point linked to the official text.

By Rhinehart Studio · Checked against official sources on 31 October 2026 · General information, not legal advice.

This guide is for small businesses and agencies running paid social, search and shopping ads to consumers in the UK, the US or both. It covers AI-generated or AI-edited images, video and voices. It is general information, not legal advice, and it doesn't cover political or election ads, which have extra rules on every platform.

1. The rule underneath everything: don't mislead

UK. The UK Code of Non-broadcast Advertising (the CAP Code), applied by the Advertising Standards Authority (ASA), has no AI-specific rules. The ASA says the Code is media-neutral, so an ad is judged on how it is likely to be understood, not on how it was made (ASA: AI and deepfakes). The rules that matter most:

  • Rule 3.1: ads must not materially mislead, or be likely to (CAP Code section 3).
  • Rule 3.3: don't leave out or obscure material information.
  • Rule 3.7: hold documentary evidence for objective claims before the ad runs.
  • Rule 3.11: don't exaggerate the capability or performance of a product.

The ASA is also clear that you are responsible for what an AI tool produces for you. In its deepfakes article it points to a ruling in which an advertiser's use of Google's automated ad tools did not shift responsibility for the resulting ad (ASA: AI and deepfakes).

US. Section 5 of the Federal Trade Commission Act declares “unfair or deceptive acts or practices in or affecting commerce” unlawful (15 U.S.C. 45(a)(1)). There is no separate federal AI-advertising statute. The same truth-in-advertising test applies: what impression does the ad give a reasonable consumer, and can you back it up? The FTC's Endorsement Guides (16 CFR Part 255) and its Consumer Reviews and Testimonials Rule (16 CFR Part 465) apply those principles to people, reviews and testimonials, and both matter for AI ads (see section 4).

2. Do you have to label AI in the UK?

Not as a blanket rule. In May 2025 the ASA said there is no general UK legal requirement to disclose AI use in ads. It suggests two questions instead (ASA: Disclosure of AI in advertising):

  1. Would people be misled if you didn't say AI was used? An obviously fantastical scene, such as a reindeer delivering your parcel, is unlikely to mislead anyone.
  2. Does your disclosure clarify the ad, or contradict it? A small “AI-generated” note can't fix an ad whose main message is false.

The ASA's own example is a cosmetics ad that uses AI to exaggerate a product's effect: a disclosure won't stop it being misleading. Disclosing an AI influencer or an obviously comedic deepfake, by contrast, can help people understand what they are looking at (ASA: Disclosure of AI in advertising). Our UK guide to labelling AI-generated ads goes into this in more detail.

3. The beauty-filter precedent and AI product shots

The closest UK precedent for AI imagery is the ASA's work on social media filters. In 2021 it upheld a complaint about an Instagram story for Skinny Tan in which a “Perfect Tan” filter was applied. It ruled that the filter was likely to exaggerate the effect the product could achieve, so the ad breached rules 3.1 and 3.11 (ASA ruling on Skinny Tan Ltd). The ASA was careful to add that filters aren't a problem in themselves. The issue is using them in a way that is directly relevant to the claim being made, for example how a tan, a foundation or a lash serum performs (ASA: the (mis)use of beauty filters; ASA AdviceOnline: production techniques).

AI image and video tools apply the same logic with far more power. In July 2026 the ASA upheld a ruling against Cablelinker Electronics, trading as Aquoxis, over online product images that appeared to be at least partly AI-generated. The images were inconsistent with each other and with the real product, and one showed components that would have blocked the water flow the product needed to work (ASA ruling on Cablelinker Electronics Ltd).

What this means for a Christmas campaign:

  • Usually fine: an AI-generated backdrop such as a fireside, a snowy street or a dressed tree behind a real product photo, or stylised festive scenes nobody would take literally.
  • Risky: AI that changes the product itself, for example its size, colour, finish, contents, the number of items in a gift set, the quality of a print, or the “after” in a before-and-after.
  • Risky: AI-generated packaging, gift boxes or bundles that differ from what will actually arrive in December.

A simple test: if a customer opened the parcel on Christmas morning, would the product match the ad?

4. AI people: presenters, “customers”, reviews and testimonials

AI presenters and avatars are popular for gift-guide videos. The law is less concerned with the avatar than with what it implies.

UK. Rule 3.47 of the CAP Code says marketers must hold evidence that a testimonial or endorsement is genuine, unless it is obviously fictitious, and rules 3.44 and 3.45 deal with fake and incentivised consumer reviews (CAP Code section 3). Submitting, commissioning or publishing a fake consumer review is also one of the practices banned outright by the Digital Markets, Competition and Consumers Act 2024 (DMCC Act 2024, Schedule 20, paragraph 13). An AI “happy customer” describing results that no real customer reported is a fake testimonial however it is labelled. An AI presenter who simply explains your product, without posing as someone who has used it, is far less likely to be a problem.

US. The FTC's Endorsement Guides were updated in 2023 (88 FR 48092). They say:

  • endorsements must reflect the endorser's honest opinions and experience (16 CFR 255.1(a));
  • if an ad shows consumers in audio or video, they should be actual consumers, or the ad should clearly and conspicuously disclose that they are not (16 CFR 255.2(c));
  • you need substantiation for the experience a consumer endorsement implies (255.2(a)), and you must not distort or misrepresent what consumers think of your product (255.2(d));
  • material connections between you and an endorser must be clearly disclosed (16 CFR 255.5).

The Consumer Reviews and Testimonials Rule took effect on 21 October 2024 (89 FR 68034). It bars businesses from writing or creating a review or testimonial that misrepresents that the reviewer exists, actually used the product, or had the experience described. It also bars spreading one that you knew, or should have known, was misrepresented in that way (16 CFR 465.2). An AI-generated “customer” giving a testimonial is exactly what this rule targets. Its definition of a celebrity testimonial includes depicting a celebrity's likeness (16 CFR 465.1).

New York's synthetic performer law. Since 9 June 2026, New York General Business Law section 396-b has required anyone who produces or creates an ad for their goods or services to disclose conspicuously in the ad that it contains a “synthetic performer”, where they have actual knowledge of it (New York Governor, 9 June 2026). A synthetic performer is an AI or software-created asset intended to look like a human performer who isn't recognisable as any real person. The civil penalty is $1,000 for a first violation and $5,000 for each later one, with exceptions for audio-only ads and AI used only to translate a human performer (N.Y. Gen. Bus. Law 396-b; bill S8420-A). If an AI presenter appears in ads shown in New York, label it.

5. Real people's faces and voices: likeness and publicity rights

A deepfake or sound-alike of a real person is the highest-risk use of AI in an ad.

UK. The ASA says a deepfake that falsely suggests a celebrity endorses your product is likely to breach the Code. It also notes that personality rights and intellectual property are outside its remit, so you should take legal advice (ASA: AI and deepfakes). Rule 6.1 of the CAP Code urges you to get written permission before portraying a member of the public, referring to someone with a public profile, or implying their personal approval. It warns that people who don't want to be associated with your product may have a legal claim (CAP Code section 6). English law has no general “image right”, but the Court of Appeal has confirmed that using a famous person's image in a way that falsely suggests endorsement can be passing off (Fenty v Arcadia [2015] EWCA Civ 3).

US. Rights of publicity are mainly state law. Two examples:

  • California: knowingly using another person's name, voice, signature, photograph or likeness in advertising without prior consent creates liability for damages (Cal. Civ. Code 3344).
  • New York: using a living person's name, portrait, picture, likeness or voice for advertising without written consent is a misdemeanour, and the person can sue for an injunction and damages (N.Y. Civ. Rights Law 50; section 51).

The practical rule is the same in both countries. Don't use AI to recreate a real person, a celebrity look-alike or a recognisable voice unless you have their written permission. A generic AI Santa is fine. A Santa who looks or sounds like a famous actor is not.

6. Platform rules: Meta, Google and YouTube

Platform policies apply on top of the law, and they can reject ads that are perfectly legal.

Meta (Facebook and Instagram). Meta applies an “AI info” label to ads created or significantly edited with its own generative AI tools, and to ads made with third-party AI tools that it detects through industry-standard signals. Minor edits such as resizing or colour correction aren't labelled. The label usually sits in the ad's three-dot menu, but where its tools add an AI-generated photorealistic human it appears next to “Sponsored” (Meta Help: AI-generated images in ads; Meta: GenAI transparency for ads). Ads about social issues, elections or politics must self-disclose AI. For organic posts, Meta requires people to disclose photorealistic video or realistic audio that was digitally created or altered, and may apply penalties if they don't (Meta, February 2026).

Google Ads. Google says all ad content is subject to its policies however it was made, and generative AI content gets the same review and enforcement. Images generated by Google Ads carry open-standard markup and an invisible SynthID watermark (Google Ads Help). Its misrepresentation policy covers unreliable claims, including improbable results, and falsely implying a relationship with or endorsement by another person or brand (Google Ads misrepresentation policy). It also says plainly: “Manipulating media to deceive, defraud or mislead others is not allowed”. Google gives at least seven days' warning before suspending an account under that policy (Google: manipulated media).

Google Merchant Center (Shopping ads). Images created with generative AI must carry the IPTC DigitalSourceType “TrainedAlgorithmicMedia” metadata, and you must not strip it out. AI-written titles and descriptions belong in the structured_title and structured_description attributes, with the digital source type set to trained_algorithmic_media (Merchant Center: AI-generated content). Many export and compression tools strip metadata, so check your files before you upload Christmas product images.

YouTube. Creators must disclose realistic AI-generated or AI-altered content, such as a real person appearing to say something they didn't, or a realistic scene that never happened. Beauty filters, lighting changes and AI help with scripts or thumbnails don't need disclosure. YouTube may add a label automatically, for example when a video carries C2PA metadata (YouTube Help: disclosing GenAI content). For video specs on each platform, see our vertical video ad specs guide.

7. Christmas-specific traps

  • Delivery promises in the creative. An AI scene of a gift under the tree on 25 December implies it will arrive in time. Make sure your cut-off dates are real and visible.
  • Stock and bundles. Don't generate a gift set, colourway or bundle you can't ship.
  • Prices and offers. AI doesn't change pricing law. Our UK sale pricing checklist and US holiday sale guide cover “was” prices, timers and fees.
  • Children's products. Ads aimed at children have extra rules in both countries, so take more care with AI characters and claims in toy ads.

8. What our AI Ad Studio does, and doesn't do

We make short AI-generated video ads for Meta Reels, TikTok and YouTube Shorts. Here's how we work within the rules above:

  • Real product, never warped. We build ads around your own product photos and footage. The video model isn't allowed to change the product.
  • No fake people or fake reviews. Every presenter is AI-generated, never a real actor or customer, and never presented as a customer. We don't imitate real people, celebrities or voices, and we don't write testimonials.
  • Your claims, your evidence. We only use claims you supply and can substantiate. We don't promise sales or return on ad spend.
  • Labelled. Our ads are labelled as AI-generated. We keep C2PA metadata where our tools provide it, and show you how to switch on each platform's AI label.

What we don't do: we don't give legal advice, clear rights in third-party material you send us, run or pay for your ad spend, or guarantee that a platform will approve an ad. AI-generated footage may not be protected by copyright, so you get a licence to use the ads rather than ownership of the footage. The 5-Ad Pack is delivered within 7 working days of script approval. Allow for that, plus platform review time, when planning Christmas campaigns.

Pre-launch checklist for AI ads

  • The product shown is your real product: same size, colour, finish, contents and quantity as what ships
  • AI was used only for backdrops, styling or obviously fantastical scenes, not to improve how the product performs
  • Any before-and-after or result shown is real and unfiltered, and you hold evidence for every objective claim
  • No AI “customers”, reviews or testimonials; AI presenters don't claim to have used the product
  • No real person, celebrity look-alike or recognisable voice without written permission
  • Ads featuring an AI human that run in New York carry a clear synthetic performer disclosure
  • Where a viewer could be misled without it, the ad says AI was used, and the label doesn't contradict the ad
  • Meta and YouTube AI labels are switched on where required; AI images keep their C2PA or IPTC metadata
  • AI-written Merchant Center titles and descriptions use structured_title and structured_description
  • Christmas delivery cut-offs, stock and prices shown in the creative are true
  • Paid partners and affiliates are disclosed, and nothing implies an endorsement you don't have
  • A person has watched every final ad, start to finish, before it goes live

FAQ

Do I have to label AI-generated images in UK ads?

There is no blanket legal requirement. The ASA asks whether people would be misled without a disclosure, and whether the disclosure clarifies the ad rather than contradicting it. A label can't rescue an ad that misleads, for example one that uses AI to exaggerate a product's effect (ASA guidance).

Can I use an AI presenter or avatar in a US ad?

Yes, if it doesn't pose as a real customer or say things no customer has experienced. The FTC's Endorsement Guides expect actual consumers in audio or video testimonials, or a clear disclosure that they are not (16 CFR 255.2(c)). In New York, ads containing a synthetic performer must disclose it (Gen. Bus. Law 396-b).

Is an AI-generated Christmas background a problem?

Usually not, provided the product itself is shown accurately. The ASA's filter rulings focus on edits that are directly relevant to the product's effect, and its Aquoxis ruling concerned AI product images that didn't match the real product (ASA ruling).

Will Meta or Google reject an ad just because it uses AI?

Not for that alone. Both say AI content is held to the same policies as any other ad. Meta may add an “AI info” label, and Google rejects manipulated media used to deceive or mislead (Meta, Google).

Do your ads guarantee approval or sales?

No. We follow the rules on this page and label our ads as AI-generated, but platforms make their own review decisions, and we don't promise sales or return on ad spend. Our work isn't legal advice.

Official sources (checked 31 October 2026)

Laws and guidance change. Check the sources above for the latest position. General information, not legal advice. For advice on your own circumstances, speak to a solicitor (UK) or an attorney (US). Platform policies change often, so check them again before each campaign.

More guides

Get one practical marketing and compliance tip a week in the Weekly Brief, free.