Do we have to label AI-generated assets in Google Ads, and where does Google show a visible AI label?
Our answer
Mark AI-made or AI-edited assets with Google's AI label setting. Google shows a visible label only on campaigns targeting the EU, India, New York, California, South Korea or Vietnam. Add your own label for realistic AI people.
- Confidence
- Medium
- Strongest proof
- 2 controlled tests
- 11 Google rules, 2 practitioners, 3 opinions or claims
- Sources
- 12 sources
- 18 statements
- G
New here? Take the 1-minute tour
Why this is the answer
3 reasons and our judgmentThe visible label shows in only six regions. Ads with assets you mark as AI made or edited get an overlay only when campaigns target those regions (Google). Elsewhere the disclosure sits in the How this ad was made panel behind the ad's menu.
Google labels assets from its own AI tools itself, and you can't overwrite those labels. It lets you add your own label without breaking its text overlay and watermark rules (Google, PPC Land).
Labels made no significant difference to brand measures in MediaScience's controlled test of 900 US viewers. The test didn't run on Google or measure sales, and Search Engine Journal reports that Google's effect remains to be seen.
Our judgment
- Mark every asset made or meaningfully edited with AI outside Google's tools, but not routine edits. Outside the six overlay regions the label shows only in the menu panel, Google may label assets itself, and Google says not all AI content needs a label. Common Thread Collective would label anything AI meaningfully shaped.
- Add your own on-ad label to any ad with a realistic AI-made person that New York or California viewers can see, rather than relying on Google's overlay. Google ties its overlay to campaign location targeting and says its setting doesn't guarantee compliance, while Andrew Foxwell says California's law reaches anyone whose ads Californians see. This matches Labeling AI ads and AI voiceover.
- Turn off image enhancements on any campaign where you add your own label. New Performance Max campaigns are opted in by default, and Google says enhancements can crop a label.
- Expect no large brand cost from labeling, but watch cost per sale in the overlay regions. Marketing Dive reports a full-length icon scored lowest, and neither that report nor MediaScience's test ran on Google or measured sales.
Our judgment, built from Google, PPC Land, IAB (Interactive Advertising Bureau), Common Thread Collective, Andrew Foxwell, Venable LLP, Davis+Gilbert LLP, MediaScience, Marketing Dive and Search Engine Journal: our reading of these sources, not a finding from any one of them.
What the sources say
18 statements from 12 sourcesGoogle says visible AI labels may appear from November 2026 on ads bought through Google Ads and Display & Video 360 in California, South Korea and Vietnam, in response to new rules. No start day is given.
“In response to upcoming regulatory requirements in California, South Korea, and Vietnam, Google is extending its AI Labeling functionality in these regions starting November 2026.”
Google Merchant Center lets merchants label a product image or video as made or edited with AI, which adds a disclosure to the ad's "How this ad was made" panel. A detected label stops automatic image enhancements.
“If an AI label is detected, automatic image enhancements like overlay cleanup from Merchant Center Automations won’t run.”
Google says not all AI-created ad content needs a label; rules vary by region, with the EU, India and New York as examples, and election ads with synthetic content need a disclosure.
“Not all AI-created content needs to be labeled.”
Google lets advertisers label AI-made ad assets; labels show in My Ad Center everywhere and on the ad itself in 6 regions, including the EU, India, New York and California. Image enhancements can crop your own label.
“Use of the AI label setting in Google’s advertising products doesn’t guarantee your compliance with specific regulations.”
Search Engine Roundtable reports Google now requires AI content disclosure in California, South Korea and Vietnam as well as the EU, India and New York, with labels on the ad itself in some places.
“In some geographies, AI labels will also show on the ad itself.”
Google says new Performance Max campaigns are opted into image enhancements by default, such as crops, overlays and animation. Its AI label setting does not guarantee compliance with EU, India or New York rules.
“New campaigns are by default opted into image enhancements.”
Andrew Foxwell says California will require a clear disclosure on ads featuring realistic AI-generated performers from January 1, for any ad seen by Californians. New York's similar law took effect June 9, 2026.
“The law can reach anyone whose ads are seen by Californians, not just brands based there”
Davis+Gilbert reports California's SB 1050, signed September 16, 2026 and effective January 1, 2027, makes it unlawful to publish ads prominently featuring a synthetic performer without a clear disclosure.
“The new law takes effect on January 1, 2027.”
The IAB's voluntary 2026 framework asks for labels on AI-generated images and video and synthetic people that could mislead, not on routine retouching or ad copy, and cites a 31.5% click-through drop from labeling.
“regardless of subsequent human refinement, editing, or compositing”
Reed Smith says that from August 2, 2026 the EU AI Act requires a clear, visible disclosure on deep-fake images, audio or video at first exposure, and ads are unlikely to qualify for lighter rules.
“Advertising is unlikely ever to qualify.”
Venable lawyers say New York has required a conspicuous disclosure on ads using AI-generated human performers since June 9, 2026, with no set wording. Audio-only ads and translation-only AI use are exempt.
“not recognizable as any actual person”
Search Engine Journal says advertisers using AI tools outside Google must apply the disclosure themselves and are responsible for judging when it's needed. Whether on-ad labels affect behavior is unknown.
“Google says advertisers are ultimately responsible for determining when AI use requires a disclosure”
PPC Land says Google can label AI-made ads even if you never touch the setting, with no way to remove the label, and labels don't carry over between Google Ads, DV360 and CM360. Its region list is now out of date.
“An advertiser who never touches the AI label setting can still end up with a labeled ad, and there is no interface control to remove one applied this way.”
Common Thread Collective, which sells ad buying, advises labeling any asset where AI meaningfully shaped the result, and says there isn't yet enough data to say whether labels hurt performance.
“when in doubt, label it. Over-disclosure carries far less risk than under-disclosure.”
Since July 2026 Google lets advertisers put AI labels inside image and video ads without breaking overlay rules, and may label assets made with its own AI tools. Using the setting doesn't guarantee legal compliance.
“Use of the AI label setting in Google’s advertising products doesn’t guarantee compliance with specific regulations.”
Google says ads made with its AI tools will automatically show an AI disclosure in a How this ad was made panel, advertisers using other tools get a control to disclose, and some regions show a label on the ad.
“This panel will indicate if an ad was created or edited with AI.”
In a 900-person study reported by Marketing Dive, positive brand attitude was 44% with a full-length AI icon, 51% with no label, and 49-51% with text labels. 60% want disclosure when AI simulates humans.
“An AI icon elicited the worst brand attitude score, with just 44% of participants saying they had a positive attitude toward the brand.”
A Mediascience experiment with 900 US respondents found no significant difference from an unlabelled control on brand choice, ad memory or brand attitude when AI-made video ads carried a disclosure label.
“if the creative is good, disclosure does not hurt it.”
In the full answer
For readers who sign in- What to do this week3 steps
- The numbers behind it3 figures from cited sources
- Where each source standsA chart placing each source on the question
- The strongest disagreementThe best case against the answer, from a cited source
- How sure we areWhat its confidence rests on, and what would change the answer