The Short Answer
Yes, Meta changed its deepfake policy in July 2026, and no, that does not mean a fake nude of you is now stuck online. As of July 2026, Meta stopped removing AI-generated content just because it is manipulated media. Instead it labels most of it as "AI info" and leaves it up, unless the content breaks a different rule. Here is the part the headlines skipped: a nonconsensual intimate image of you, including an AI deepfake, breaks a different rule. It is not judged as "manipulated media" at all. It falls under Meta's Adult Sexual Exploitation policy, which still requires removal, and under the US TAKE IT DOWN Act, which forces covered platforms to take nonconsensual intimate imagery down within 48 hours of a valid report. So if you searched "Meta won't remove a deepfake of me" in a panic, the honest answer is that the labeling change almost certainly does not apply to your case. This guide shows you which path actually applies and how to invoke it.
What Meta Actually Changed in July 2026
Through the first half of 2026, Meta rolled out AI content labeling across Facebook and Instagram, tagging AI-generated or AI-edited media with an "AI info" label. In July 2026 the enforcement side shifted: rather than taking manipulated media down, Meta's default is now to label it and leave it up. The reasoning it gives is that a label preserves context and free expression while still telling viewers the content is synthetic.
That is a real change, and for a satirical political deepfake or an obviously edited meme, it means the content stays online with a tag. Coverage of the shift has mostly looked at it from an advertiser and general-tech angle, which is where the "Meta stops removing deepfakes" panic comes from. But labeling-instead-of-removing is a rule about manipulated media as a category. It was never the rule that governs intimate images of real people, and that distinction is the whole story for a creator.
Why Your Case Is Almost Never "Just a Label"
Meta does not run every piece of content through one giant "is this AI" filter and stop there. Content is judged against separate policies, and the harshest applicable one wins. A deepfake gets the "AI info" label under the manipulated-media policy. But a sexual or intimate deepfake of a real, identifiable person, shared without their consent, also violates the Adult Sexual Exploitation policy, and that policy calls for removal, not a label.
This is not a theory. Meta's own Oversight Board has ruled on exactly this: when the company treated explicit deepfake images of real women as a mere manipulated-media problem, the Board said that was wrong and that such content belongs under the adult sexual exploitation rules, where it must come down. So the correct mental model is this:
- A deepfake that is not sexual (a fake photo of you at an event, a manipulated interview clip) may now be labeled and left up under the July 2026 change.
- A deepfake that is intimate or sexual and was made or shared without your consent is nonconsensual intimate imagery. It is removable, and the labeling change does not shield it.
If you are a creator worried about a fake nude, an AI "undress" image, or a face-swapped explicit video, you are in the second bucket, not the first.
The Removal Path That Still Works
You have two levers, and they stack. Use both.
1. Report it as nonconsensual intimate imagery, not as a deepfake
This wording matters. If you report the post as "AI-generated" or "fake", you route it to the manipulated-media queue, the exact queue that now ends in a label. Report it instead as a nonconsensual intimate image of you. On Instagram and Facebook, use the report flow on the post and choose the options for nudity or sexual activity involving you, or "involves someone I know" and then the intimate-image option. The goal is to land the report under the Adult Sexual Exploitation policy, where removal is the required outcome.
2. Invoke the TAKE IT DOWN Act 48-hour rule
Since May 19, 2026, the US TAKE IT DOWN Act requires covered platforms, including Meta, to provide a clear removal request process for nonconsensual intimate imagery and to remove valid reports within 48 hours, including reasonable efforts to remove copies. Deepfakes are explicitly covered. When you file, state plainly that the image is an intimate depiction of you published without your consent, that it is a digital forgery if it is AI-generated, and include the exact URLs. If a platform blows the 48-hour deadline, the Federal Trade Commission enforces the Act, and there is an official complaint portal at TakeItDown.ftc.gov. Our full guide to the TAKE IT DOWN Act walks through the request step by step.
Label vs Remove: Who Does What in 2026
Platforms are not moving in the same direction, so it helps to see them side by side. This is the current lay of the land for a creator dealing with a synthetic intimate image.
| Platform | Generic AI deepfake | Nonconsensual intimate deepfake of you |
|---|---|---|
| Meta (Instagram, Facebook) | Labeled "AI info", usually left up (July 2026 change) | Removable under Adult Sexual Exploitation policy plus 48-hour TAKE IT DOWN duty |
| YouTube | Disclosure label for synthetic media | Likeness-detection and privacy removal on request; NCII removable |
| X | Labels plus tightened synthetic-media rules from early 2026 | Nonconsensual nudity is prohibited and removable on report |
| Search engines (Google, Bing) | Not their call to label | De-index the URL via the involuntary fake-imagery and copyright removal forms |
The takeaway: the labeling debate is about the generic column. Your case lives in the removable column on every one of these surfaces.
Step by Step: Getting a Fake Nude of You Off Instagram or Facebook
- Capture evidence first. Screenshot the post, the profile, and the URL before you report. If it is removed, you still want the record, and identical copies may surface elsewhere.
- Report as an intimate image, not as a fake. Use the in-app report flow and choose the nudity or nonconsensual intimate image path, not the "false information" or "AI" path.
- File a TAKE IT DOWN request in parallel. State it is a nonconsensual intimate depiction of you, note if it is a digital forgery, and list every URL. Keep the confirmation and timestamp.
- Hit the search engines too. Even after the post is gone, the link can sit in search results. Google's removal process covers involuntary fake explicit imagery and copyright; de-indexing keeps people from finding it.
- Watch for reuploads. Nonconsensual images rarely appear once. The same file gets re-posted, mirrored, and re-indexed for weeks. Re-check, and re-report each copy the same way.
If the material is content you created, copyright is a third, clean lever that works regardless of how the platform classifies the image. A DMCA notice obliges hosts and search engines to act because you own the work. Our complete DMCA guide covers that route.
When This Outgrows Doing It Yourself
For a single post on one platform, the steps above are enough, and you should not pay anyone to do what a report form does for free. The threshold changes when the same fake is spreading: reposted across accounts, mirrored onto adult tube sites and leak forums that have no report button, and reappearing every time you clear it. At that point the work stops being paperwork and becomes continuous discovery, host takedowns that actually get answered, coverage across every search engine, and monitoring for reuploads, which is the work a removal service exists to do. If you are not sure which situation you are in, run a free scan and see how far the content has actually spread. If it is one link, handle it yourself with the steps above. If it is fifty, now you know why it felt impossible by hand.
This article is educational and not legal advice. For your specific situation, a lawyer beats a blog post, ours included.
