Meta's Oversight Board wants AI deepfakes removed and monetization tightened. Only part of that binds Meta.

What happened

Meta's Oversight Board has called on the company to remove AI deepfakes and to tighten its content policies around synthetic media, including how that content is monetized.

The board is the external review body Meta set up to rule on contested moderation decisions. It does two different things, and the difference matters this week. It decides individual cases, and it issues policy recommendations.

Only the case decisions bind Meta. The policy recommendations are advice the company can accept, partially accept, or decline.

That is the whole story for anyone trying to figure out what changes on their account.

What this is not

It is not a new Meta policy. Nothing in your Instagram, Facebook, or Reels rules changed the moment the board spoke. Meta typically responds to recommendations on its own timeline, and "we are assessing this" is a valid response.

It is not a ban on AI-generated content. The pressure described in the reporting is about deepfakes — synthetic depictions of real people — and about who gets paid for that material. Making a video with an AI editor is not the target. Making a video of a real person who never said the thing is.

It is not a US law. Separately, courts have struck down deepfake statutes on speech grounds, while the EU has been building a transparency code of practice around marking AI-generated content. Platform rules and statute are moving on different tracks and at different speeds. Right now the platform track is the one that can demonetize you next month.

Why creators should plan for the strict version

The monetization piece is the part worth reading twice. Removal policy affects the person posting the deepfake. Monetization policy affects everyone who uses synthetic media anywhere near a real person's face or voice — including commentary channels, parody accounts, reaction content, and anyone using an AI presenter.

When platforms tighten synthetic-media rules, enforcement almost never arrives with a clean definition. It arrives as a classifier. The classifier does not know your clip is satire, and it does not know the voice in your cold open is your own voice run through a cleanup tool.

The practical risk is not a strike. It is a silent monetization hold on a video that took you three days, with an appeals process that runs longer than the video's earning window.

So the planning assumption should be the strict one: label synthetic depictions of real people, keep your source material, and do not build a recurring format that only works if Meta keeps paying for unlabeled synthetic faces.

What to do about it

1. Audit your last 30 days of Meta posts for anything that depicts a real person using AI — voice, face, or both. That is the exposure list.
2. Turn on whatever AI-content disclosure field Meta offers you and use it on those posts, even when you think it is obvious satire. Self-labeling is cheap; a demonetization review is not.
3. Keep raw files and prompts for anything synthetic you publish. If a hold lands, the appeal is faster when you can show the chain.
4. If an AI presenter or cloned voice is central to a format, get a second distribution surface running now. Do not let one platform's policy change take a format offline.
5. Watch for Meta's formal response to the recommendations. Until it publishes one, treat every headline about this as pressure, not policy.
6. Write your own house rule for depicting real people with AI, and put it in writing before a platform writes one for you.