TL;DR
DMCA takedown vs deepfake removal comes down to whether the content is real footage of you used without permission or synthetic. A DMCA notice protects copyright in a real recording and has no fixed deadline. A TAKE IT DOWN Act notice covers non-consensual intimate imagery, including AI-generated content, and requires removal within 48 hours of a complete notice. If a leaked original and deepfakes derived from it both exist, you need both, filed separately.
Two different questions, two different laws
If you've found something of yours online that shouldn't be there, don't start with "how do I get this removed." Start with "which legal mechanism applies here." The answer depends on one thing: is the content real, or synthetic?
A DMCA takedown notice and a TAKE IT DOWN Act notice look alike on the surface. Both are written requests to a platform or host, both ask for removal, and both can be filed by the person affected or someone acting for them. But they answer different questions, target different platforms, and move at different speeds. Filing the wrong one, or only one when you needed both, is one of the most common reasons a removal request stalls.
The core distinction
DMCA takedowns (17 U.S.C. §512) exist to protect copyright. The question a DMCA notice answers is: is this your footage, being used without your permission? It only works when the content is a real image or video of you, something that actually happened and was actually recorded.
The federal TAKE IT DOWN Act (Public Law 119-12, signed May 19, 2025) protects people from non-consensual intimate imagery, including content that was never real. The question it answers is: is this a non-consensual depiction of you? It doesn't matter whether the event ever happened. AI-generated "digital forgeries" are explicitly covered, which is the gap copyright law leaves open: you can't claim copyright over a video that was never filmed.
The one-sentence version
DMCA asks whether you own the copyright in a real recording. The TAKE IT DOWN Act asks whether a real person is depicted non-consensually, whether or not the footage is real.
Side-by-side comparison
| DMCA Takedown | TAKE IT DOWN Act Notice | |
|---|---|---|
| Legal basis | 17 U.S.C. §512 (copyright) | Public Law 119-12 (federal, 2025) |
| What it covers | Genuine copyrighted content used without permission | Non-consensual intimate imagery, including AI-generated deepfakes |
| Core question | Is this your real recording? | Is this a non-consensual depiction of you? |
| Requires real underlying footage | Yes | No — synthetic content qualifies |
| Response deadline | No fixed deadline in the statute | 48 hours from a complete, valid notice |
| Who can file | Copyright holder or authorized agent | The depicted individual or an authorized representative |
| What a valid notice needs | Identification of the work and infringing material, good-faith statement, signature | Signature, precise URL(s), good-faith statement it's non-consensual, contact info |
| Applies to | Any host/platform storing the infringing copy | "Covered platforms" — public-facing services primarily hosting user-generated content |
How to figure out which one applies to you
The legal options decision tree walks through this branching logic interactively, or you can step through it yourself:
- Is the content real footage of you, something that was genuinely filmed or photographed, even if it was meant to stay private? If yes, DMCA applies, because you (or whoever holds the copyright, often the creator) own the underlying work.
- Is the content AI-generated, manipulated, or otherwise synthetic, such as a face-swap, a generated video, or anything a reasonable viewer might take for real but that shows something that never happened? If yes, the TAKE IT DOWN Act applies and DMCA generally does not, because no genuine copyrighted work is being infringed.
- Is it intimate or sexual in nature? The TAKE IT DOWN Act specifically covers non-consensual intimate imagery. A deepfake that isn't intimate (say, a fabricated video of you saying something you never said, used for a scam) falls outside this statute, though it's still worth reporting to the platform directly. The TAKE IT DOWN Act eligibility checker can confirm this for a specific case in under a minute.
Sending the wrong notice slows you down
Filing a DMCA claim against a deepfake usually gets rejected or ignored, because there's no copyrighted work to point to. Filing a TAKE IT DOWN Act notice against a genuine leaked photo can also get flagged as the wrong route, since it isn't framed as a copyright claim a host's DMCA agent recognizes. Matching the notice to the content type isn't a formality; it determines whether the platform can act on it at all.
The case where both apply
A common trap: a real photo or video of you leaks, and then someone uses that material as a source to generate deepfakes, by face-swapping you into other footage or using it to prompt an AI image generator. Here you're usually dealing with two separate pieces of content that need two separate notices:
- The original leaked photo/video is a genuine copyrighted work of you. File a DMCA notice against every URL hosting that original file.
- The AI-generated content derived from it is a digital forgery under the TAKE IT DOWN Act, even though it started from something real. File a TAKE IT DOWN Act notice against those URLs separately, since there's no copyright claim to make against synthetic output.
Treat these as two parallel tracks, not one combined case. They may go to different platforms, need differently worded notices, and carry different response-time expectations: the deepfake notices have a hard 48-hour clock once complete, while the original-photo DMCA notices have no statutory deadline at all.
Why the distinction matters beyond just "which form to fill out"
Getting this right affects whether removal happens at all, not just whether the paperwork is correct. A platform's DMCA agent is often a separate team, with separate tooling, from whoever handles TAKE IT DOWN Act or general trust-and-safety complaints. A notice routed to the wrong queue can sit unanswered indefinitely, not because the platform is ignoring you, but because nobody equipped to act on it ever reviewed it.
It's also why identity-based detection matters more than filename or keyword matching. A tool that only searches for your name in a filename won't tell you whether what it found is a real leak or a synthetic fabrication, and that classification decides which notice to file. Detection that verifies media authenticity as part of the scan, instead of treating every match the same, saves you from guessing.
What this looks like in practice
If you're handling this on your own, a workable process looks like:
- Confirm what you're looking at. Is it a real recording, or does something look off, such as inconsistent lighting, unnatural movement, or a face that doesn't quite track? When in doubt, treat unclear cases as deepfakes and use the TAKE IT DOWN Act route, since its safe-harbor provision protects platforms that remove content in good faith even if it later turns out otherwise.
- File the matching notice per URL, not per site. A domain-level complaint isn't enough for either law, and both expect precise locations.
- Track response times separately. A TAKE IT DOWN Act notice that hasn't led to removal within 48 hours has grounds for escalation (the FTC treats a platform's failure to reasonably comply as an unfair or deceptive practice). A DMCA notice has no equivalent clock, so escalating there means re-filing, contacting the host directly, or pursuing de-indexing in parallel.
- Re-check for reuploads. Leaks and deepfakes both tend to reappear after removal, because the same source material or the same generation prompt produces new copies. Whatever process caught the content once should keep watching the same identity.
If sorting out which notice applies to a growing list of findings is more manual triage than you want, Noticeora's monitoring handles that classification. Every match gets a type (leak or deepfake), the claim type follows from it, and Autopilot files the correct notice without you doing the legal analysis case by case. A free scan shows what turns up before you decide how hands-on you want to be.
