Google has clarified that an AI disclosure does not make an ad acceptable when the ad uses AI to impersonate a public figure. The November 11 update to Google Ads’ Unacceptable Business Practices policy makes that point explicit, while Google says it is clarifying the existing rule rather than changing enforcement. For marketers using generative tools to scale creative, the practical lesson is clear: labeling an asset as AI-generated is not a substitute for permission, truthful endorsement claims, or review.
What Google clarified, and what it did not
Google’s policy notice says ads that impersonate public figures using AI are not allowed “even if they disclose that they were generated or edited using AI.” The company says the clarification is intended to help advertisers understand the policy and “does not change the enforcement of the policy.” The notice was posted November 11, and industry coverage followed the next day.
That distinction matters. This is not a newly announced grace period or a new enforcement start date. It is a sharper answer to a question creative teams may have had: if an ad is labeled as synthetic, can it still imply that a real person endorses a product? Google’s answer is no. An AI label describes how an asset was made; it does not establish that the person agreed to appear in it.
Read the full clarification in Google’s November 2026 policy update and the contemporaneous summary from Search Engine Roundtable.
The risk is false endorsement, not every use of AI
The rule sits under public-figure impersonation, not a blanket ban on AI-created advertising. Google’s examples focus on deception: using a famous singer’s image to suggest they are a satisfied client, falsely claiming an athlete sponsors a product, or creating a video that misleads viewers into believing a popular creator endorsed an offer. The policy’s definition of public figures includes entertainers, athletes, online influencers, government officials, and prominent business people such as well-known founders or CEOs.
Google’s account-suspension guidance says advertisers should have consent before using a public figure’s name, image, video, or voice to suggest an endorsement and should be able to prove it. It lists contracts, written agreements, credible reporting, and public statements as possible evidence. The November clarification does not detail every licensed, parody, or transformative use, so teams should not assume a disclosure alone resolves an ambiguous case.
For marketers, this is a rights-and-claims control problem. A campaign can cross the line through the combination of a synthetic face or voice, copy that implies sponsorship, and a landing page that repeats the claim, even if each asset was generated by a different tool or vendor.
Why a creative mistake can become an account-level problem
Google classifies public-figure impersonation as an egregious violation. Its guidance says accounts found in violation may be suspended upon detection without prior warning, and advertisers may not be allowed to advertise again. Appeals are possible, but Google says reinstatement is reserved for compelling circumstances. These are existing enforcement terms, not penalties newly created by the November notice.
That raises the stakes for teams using automated creative expansion, external agencies, creator marketplaces, or AI tools that turn a prompt into multiple image, video, and copy variants. An approval of the original concept may not cover every generated variation. The operational lesson aligns with our analysis of AI content quality and trust: greater production speed only helps when review and ownership controls keep pace.
A practical audit for AI-assisted ad teams
Before launching or refreshing campaigns, review the full path from source asset to final destination:
- Inventory current and queued creative. Flag recognizable public figures, synthetic voices, lookalikes, fake testimonials, and wording that could imply a real endorsement.
- Verify rights and claims. Keep documented consent tied to the specific person, asset, channel, usage period, and endorsement claim. A generic AI label is not evidence of consent.
- Gate generated variants. Require human approval for AI-expanded text, images, video, and landing-page changes. Store the approved version, rights records, and reviewer so teams can trace what actually ran.
- Escalate uncertainty before spend. If a campaign depends on a public figure’s likeness, endorsement, parody, or a synthetic alteration, pause the launch and get qualified policy or legal review rather than testing the interpretation in-market.
AI can still make campaign production faster, but the responsibility for the representation remains with the advertiser. Real Internet Sales helps businesses build accountable digital marketing systems, from creative review to search strategy. Call 803-708-5514 or visit realinternetsales.com to discuss how to scale marketing without losing control of your brand.