AI disclosure in advertising, platform by platform

Do you have to disclose AI in ads?

On our reading, yes for a fully synthetic AI actor on TikTok, in New York and, on the Commission’s guidelines, in the EU. YouTube asks at upload about AI content that “seems realistic”, ad videos included. Meta and Google Ads ask only in political or election ads, so neither has a column.

The EU guidelines leave AI-written ad copy outside, bar claims on, for example, “health, consumer safety or sustainability”.

The IAB’s framework exempts “Generic synthetic voices” that “don’t impersonate specific individuals”. It tells advertisers to “yield to regulation wherever binding requirements are stricter”.

Meta requires AI disclosure only in political and social issue ads

Meta has required political and social issue advertisers to disclose AI “in certain cases” since January 2024 (Meta, 5 April 2024).

Since 2024, Meta has labelled ads its own generative AI features created or significantly edited, its post of 3 February 2025 says. For an AI-generated “photorealistic human”, the post said the label “will appear next to the Sponsored label”. Whether that still holds since its update of 1 June 2026 is unclear.

Meta’s AI disclosure rules quotes each case, boosted posts and optional self-disclosure included.

TikTok requires a label or disclaimer on significantly AI-made ads

TikTok’s ads policy says “Significantly edited media and AIGC content is allowed” if you “Apply the AIGC label, or” add “a clear disclaimer, caption, watermark, or sticker of your own”. If TikTok finds undisclosed AI content, “your ad will be rejected or restricted” (section “Edited media and AI-generated content (AIGC)”, updated April 2026).

The policy’s examples of significant edits include “Content that contains images, video, or audio that are completely AI-generated”. They also include making the primary subject “say something they didn’t actually say, using AI voice-cloning”.

“Insignificant AI edits” are “minor tweaks or enhancements”, such as “Removing or modifying backgrounds”. The policy names no label for them, but “If you are uncertain” it says to “use caution and label it accordingly”.

TikTok’s ad disclaimers page (updated September 2025) lists AI media under “Mandatory disclaimers”. Its “AI-generated content” disclaimer type shows “at the bottom of the in-feed video for the duration of the ad”.

TikTok’s AI-generated content label for ads covers Spark Ads, organic posts and TikTok’s own labels.

Google Ads asks in election ads; YouTube asks at upload

Google’s political content policy says advertisers “must disclose all election ads that contain synthetic or digitally altered content” by ticking “Altered or synthetic content”.

For other ads, Google’s policy update of 9 July 2026 says “AI regulations in the European Union, India, and New York require that ads with certain AI-generated or edited assets include disclosures and/or labels”. Advertisers “can add these labels directly to their creatives or use the AI label setting”.

Google’s help page on the setting says labelled assets get a note in “How this ad was made”, in the three-dot menu. “For campaigns that target the European Union, India, and New York”, ads also carry “visible overlays”.

YouTube’s help page requires creators to disclose AI content that makes “a real person appear to say or do something they didn’t do”. Altered real footage and “a realistic scene that didn’t actually occur” need disclosure too.

Google Ads Help says that when you upload an ad video to YouTube from Google Ads, you are asked whether AI was used, “to comply with YouTube’s AI disclosure policies for organic videos”.

New York’s synthetic performer law has applied since 9 June 2026

New York’s S8420-A, signed 11 December 2025, amends General Business Law 396-b. The Governor announced on 9 June 2026 that the law “is now in effect”.

A synthetic performer is “a digitally created asset” made with generative AI or “a software algorithm”. It is meant to create the impression of “an audiovisual and/or visual performance of a human performer who is not recognizable as any identifiable natural performer” (1(c)).

The law does not define “recognizable”, so whether a licensed replica of a real actor fits is unclear.

Under subdivision 3, the duty falls on “Any person engaged in the business of dealing in any property or service” who “for any commercial purpose produces or creates” an ad for it. It “shall conspicuously disclose in such advertisement that a synthetic performer is in such advertisement, where such person has actual knowledge”.

Penalties are $1,000 for a first violation and $5,000 for each one after (subdivision 3).

The law “shall not apply” to “Audio advertisements” (7(a)) or where AI “solely involves the language translation of a human performer” (7(b)). Ads for expressive works such as films are exempt where the performer’s use is “consistent with its use in the expressive work” (4). The media that publish ads are outside it (8).

Amazon requires sellers to tag AI-generated people, CNBC reported

The tag covers listing images and videos, CNBC reported on 23 July 2026, quoting Amazon: “Recent legislation requires disclosure when images or videos in advertisements contain photorealistic AI-generated people.”

California’s SB 1050 covers the voice

California’s SB 1050 makes it unlawful “to create and cause to be published in an advertising medium an advertisement that prominently includes a synthetic performer without a clear and conspicuous disclosure” (17610(b)).

Its synthetic performer is “a digital figure, voice, or representation created in whole or in part using generative artificial intelligence that creates the realistic impression of the audio, audiovisual, or visual performance of a human performer who is not recognizable as any identifiable natural person” (a)(6).

“Prominently” covers a performer “In the foreground and demonstrating or illustrating the product or service” (a)(5)(A), or “Providing or voicing the on- or off-camera narration or commercial message” (a)(5)(B). A performer “Illustrating or reacting to” that narration or message counts too (a)(5)(C).

SB 1050’s “advertisement” includes any “digital communication, or other representation” (a)(1), but its advertising medium makes available “an audio or audiovisual advertisement” (a)(2)(A). Whether SB 1050 covers a still-image ad is unclear.

On our reading, (a)(2)(A) brings in audio ads, which New York’s law exempts.

The disclosure must use wording “substantially similar to” “this performance features a synthetic performer” or “no human performer is depicted” (c). The exemptions match New York’s for expressive works and translation, and add “other accessibility features” (d)(5).

Unlike New York’s, California’s law reaches platforms: an “online platform” is an “advertising medium” (a)(2)(A). Once served with a court order and information identifying the ad, the medium must, “as soon as commercially reasonable and technically feasible”, remove it in California and “Cease accepting payment for further dissemination of the advertisement” (e).

The text states no operative date. Under the state constitution (Article IV, 8(c)(1)), a regular-session statute starts “on January 1 next following a 90-day period from the date of enactment”: 1 January 2027, on our count.

The EU AI Act, Article 50: deep fakes now, older generators’ marks by 2 December 2026

Article 50 of the EU AI Act, Regulation (EU) 2024/1689 (EUR-Lex), has applied since 2 August 2026.

Under 50(4), “Deployers of an AI system that generates or manipulates image, audio or video content constituting a deep fake, shall disclose that the content has been artificially generated or manipulated.” For “evidently artistic, creative, satirical, fictional” works, the duty shrinks to a disclosure that does not hamper the work.

A deep fake is content that “resembles existing persons, objects, places, entities or events and would falsely appear to a person to be authentic or truthful” (3(60)).

Under 50(2), providers of AI systems “generating synthetic audio, image, video or text content” must ensure outputs are “marked in a machine-readable format and detectable as artificially generated or manipulated”. The duty lapses where a system performs “an assistive function for standard editing” or does not “substantially alter” the input.

The Digital Omnibus (Regulation (EU) 2026/1744) gives generators “placed on the market before 2 August 2026” until 2 December 2026 (new Article 111(4)).

Fines for breaching Article 50 go up to 15 million euros or 3 per cent of worldwide annual turnover, “whichever is higher” (99(4)(g)). For SMEs, and since the Digital Omnibus for small mid-caps, the lower of the two applies (99(6) and 99(6a)).

The Commission’s Article 50 guidelines count two AI UGC formats as deep fakes

The Commission’s Article 50 guidelines (20 July 2026) apply only once “formally adopted” in every language, its covering communication says.

Both sit outside the artistic and fictional carve-out. One is “a realistic synthetic influencer testing out a sponsored real product, focused solely on displaying the product functionalities”. The other is a teleshopping-style ad with “simulated consumers”.

The same guidelines say “A real product (e.g., a car) shown in an advertisement against an AI-generated background” is no deep fake, “as long as the ad is not likely to mislead” about the product. An AI product image that can “mislead as to the actual product appearance, characteristics or use” is listed as one.

The guidelines name “an advertising company” as a deployer. A company “remains a deployer even if it involves third parties (e.g. contractors, freelancers)” under its control (paragraph 14). A client that “merely commissions” the ad, “without taking decisions and exercising control” over the agency’s use of AI, “is not a deployer”.

A deployer outside the EU is covered when it directs content into the Union, “including by posting deep fakes on the globally accessible internet” (paragraph 13).

The Code of Practice is voluntary, and an EU icon alone does not establish compliance

The Commission’s Code of Practice, published in final form on 10 June 2026, is voluntary. The Commission and the AI Board have confirmed it as “an adequate voluntary tool to demonstrate compliance”.

Using one of the EU icons, including “Fully AI-Generated” and “Partially AI-Modified”, “does not establish legal compliance by itself”.

The Commission’s placement summary asks for the icon “directly embedded into the deepfake or published text (except for creative works), unless equivalent alternatives are available such as a user interface overlay”.

No text we read says whether a platform label meets a law

Google’s policy update of 9 July 2026 says use of its AI label setting “doesn’t guarantee compliance with specific regulations”.

New York asks for a conspicuous disclosure “in such advertisement” and sets no wording. “In most instances”, Meta’s Business Help Center puts the AI info label behind the three-dot menu, in About this ad. Whether that meets New York’s text is unclear.

TikTok’s disclaimer and Google’s overlays show on the ad itself. On our reading, that is the kind of “user interface overlay” the EU’s icon summary names as an alternative to an embedded icon. Whether either meets Article 50(4) is unclear.

California asks for wording “substantially similar to” its two sentences. TikTok describes its AI disclaimer as “a textual disclosure label indicating that the ad includes AI-generated content”. Whether that wording is substantially similar is unclear.

Where a tool’s own terms require a mark or label, the “Marks and labels” rows of the AI commercial use matrix show it.

The FTC’s testimonial rule bars fake reviewers, with no AI label

The FTC’s fake-testimonial rule, 16 CFR 465.2, has applied since 21 October 2024. Are AI UGC ads legal? applies it to AI actor scripts, clause by clause.

AI ad disclosure dates, December 2025 to January 2027

Two AI ad disclosure dates are still ahead: 2 December 2026 for older EU generators and 1 January 2027 for SB 1050, on our count.