The label is not a loophole anymore
For the last two years, a quiet belief spread through the AI content world: if you labelled it, you were safe. Put “AI-generated” in the corner, disclose it in the caption, and a video of a famous face selling a product nobody endorsed was somehow just marketing.
On 11 November 2026, Google closed that door in writing. In an update to its Unacceptable Business Practices policy, Google clarified that ads which impersonate public figures using AI are not allowed — even if they disclose that they were generated or edited using AI. Google says the update is a clarification, not a change in enforcement. In plain language: this was always the rule; now nobody can claim they didn’t know.
This is the most important AI policy story of the month for anyone who creates, advertises or earns with AI characters — and it points directly at where the safe, profitable side of this industry actually is.
What exactly did Google change?
The clarification sits inside the Google AI impersonation policy on public figures, one of the most serious categories in its advertising rules. Violations of Unacceptable Business Practices are classed as egregious, which means an advertiser account can be suspended on detection, without a warning cycle.
Three points matter most:
1. Disclosure does not cure impersonation. An AI label explains how an ad was made. It does not create consent, and it does not make a false endorsement true. A disclosed fake is still a fake.
2. The test is the impression, not the technique. The policy is not only about sophisticated deepfakes. A doctored photo, an edited clip or an AI voice that makes it look or sound as if a public figure is involved can fall under the same rule.
3. Consent is the dividing line. Google’s guidance points advertisers toward documented permission before using a public figure’s name, image, video or voice. No permission, no campaign — whatever the label says.
Why Google is moving now
This clarification did not arrive alone. It lands in a season of pressure on every major platform: investigations this month reported hundreds of ordinary-looking apps hiding abusive deepfake features behind harmless descriptions, regulators in Europe are enforcing deepfake disclosure duties, and Denmark has tabled its own deepfake legislation. Platform trust teams are responding the same way everywhere — by narrowing the grey areas that bad actors hide inside.
For Google specifically, impersonation is also an advertising-integrity problem. Fake celebrity endorsements are one of the oldest scam formats on the internet; AI simply made them cheap, fast and convincing. A rule that says “the label doesn’t save you” removes the cheapest defence those campaigns had.
What this means if you create AI characters
If your work is built on fictional characters — people who do not exist, created from text prompts, owned entirely by you — this update is not a threat. It is a competitive advantage.
The creators exposed by the new clarification are the ones borrowing real faces: fan accounts monetising a celebrity’s likeness, advertisers faking endorsements, pages trading on lookalikes. Every enforcement wave against that model pushes audience and advertisers toward original characters, because an invented character carries no consent problem, no takedown risk and no policy cliff-edge.
There has never been a better moment to do the unglamorous work properly: build a character from scratch, keep her consistent with a character sheet, give her a personality and a world, and let her become the face people recognise — her own face, not somebody else’s.
What advertisers and affiliates should do today
If you run ads or promotions anywhere near AI-generated people, use this checklist before your next campaign:
- Audit every creative for real-person likeness: faces, voices, names and implied endorsements.
- Remove borrowed fame — no celebrity lookalikes, no “as seen with” framing, no fake quotes.
- Keep consent paperwork for any real person who does appear, and be able to produce it.
- Disclose AI generation anyway. Disclosure no longer saves an impersonation, but honest labelling still matters for trust and for the rules that do require it.
- Shift budget to original characters. A fictional character you control is an asset; a borrowed face is a liability with a suspension attached.
The bigger picture for AI content in 2026
Read alongside Google’s recent search guidance — which stresses original, manually reviewed, genuinely useful content — the direction is unmistakable. The platforms are not turning against AI. They are turning against deception at scale: fake people endorsing real products, mass-produced pages with nothing original in them, content built to trick rather than to serve.
Creators who build real characters, publish real expertise and label honestly are exactly who these rules leave standing. That is the game this site plays, and this month’s clarification just made the field smaller for everyone playing it the other way.
Bottom line: as of November 2026, an AI label is transparency, not permission. Build your own characters, keep your paperwork clean, and Google’s toughest policy update of the season becomes your competitors’ problem — not yours.
For the practical side, start with our guides to creating your first fictional AI character and keeping the same character consistent in every image.
Create Your Own Fictional AI Character — 18+ Only
SoulGen AI — Create a realistic or anime fictional character from a simple text description, edit her with prompts, and reuse the same character again. Try SoulGen AI →
Seduced AI — Build reusable fictional characters and turn them into images and short videos, with more models and scene controls. Try Seduced AI →
Affiliate disclosure: We may earn a commission if you sign up through these links, at no extra cost to you. 18+ only. Create only fictional adults, or adults who have clearly given you permission. Never upload or edit a real person’s photo without their consent.