# Remove a deepfake of yourself: what to do first

Remove a deepfake of yourself: preserve evidence, report to platforms, use hash matching, ask search engines to delist and know when the law is the route.

---


To remove a deepfake of yourself, work in this order: record where it appears without copying or sharing the file, report it to every platform that hosts it as synthetic content made without your consent, ask search engines to stop listing it, and escalate to a lawyer or the police when a host ignores you or the content is sexual, threatening or used to defraud. No single request removes it everywhere, so the work is a sequence, and the order matters.

A deepfake is an image, video or audio recording that has been altered or generated by software so that it appears to show a real person doing or saying something they did not do. The harm depends on what the fake does. A sexual deepfake attacks dignity and safety. A fake endorsement or a cloned voice is used to steal money. A fabricated statement attacks your standing. Each kind has different reporting categories, a different legal footing and a different chance of success, which is why this guide starts by sorting the fake before it tells you where to click.

This guide is written for the person depicted, or for someone helping them, and it is deliberately limited to response. It says nothing about how fakes are made or where they are found. Platform and legal rules change often. Where an official page could not be read for this guide, the text says plainly that a point is left at the level of general practice. Check each live page before you file. This is information, not legal advice: your own lawyer decides what applies where you live.

<div class="box"><b>The short version</b>
<ul><li>Sort the fake first: sexual, fraud or impersonation, false statement attributed to you, or commercial use of your likeness. The route and the urgency follow from that.</li><li>Capture the page address, the date and what the page says, not the file itself, and never forward a sexual fake to anyone. If the person shown is under 18, do not copy it at all.</li><li>Platform reports remove the copy on that platform. Search removal stops it appearing in results. Hash matching stops known copies being re-uploaded on partner platforms. Each does a different job.</li><li>Copyright notices work only when you own the source material, and legal routes depend on your country and state. A lawyer is the right next step when a host refuses or the fake is being used against you.</li><li>When a route fails, the fallback is not silence: it is escalation to the host, the registrar, the police or a court, and accurate context where the fake cannot be removed.</li></ul></div>

## Which removal route fits the kind of deepfake

The route depends on what the fake does, not on how it was made. Find your row first, because it sets what to report it as and how fast to move. The table describes typical routes as general practice, not outcomes anyone can promise.

<div class="tablewrap"><table><thead><tr><th>What the fake does</th><th>Report it as</th><th>Also use</th><th>Urgency</th></tr></thead><tbody>
<tr><td><strong>Shows you in sexual or intimate content</strong></td><td>Non-consensual intimate imagery, synthetic or manipulated media</td><td>Hash matching, search removal, police, specialist lawyer</td><td>Highest: same day</td></tr>
<tr><td><strong>Impersonates you to take money or data</strong></td><td>Scam, impersonation, fraud</td><td>Your bank, the police, warnings to contacts</td><td>High: hours</td></tr>
<tr><td><strong>Puts false words or acts in your mouth</strong></td><td>Manipulated media, misleading content, harassment</td><td>Public correction, lawyer if it is false and damaging</td><td>High if it is spreading</td></tr>
<tr><td><strong>Uses your face or voice in ads or products</strong></td><td>Impersonation, deceptive advertising, likeness misuse</td><td>Ad platform reporting, a lawyer, copyright if you own the source</td><td>Medium: days</td></tr>
<tr><td><strong>Is labeled as parody or satire</strong></td><td>Harassment or impersonation if it crosses the line</td><td>Direct request to the poster, then the platform</td><td>Varies: often a hard case</td></tr>
</tbody></table></div>

Two cases deserve an early warning. If the person depicted is a minor, the matter is a child-protection and police matter first, and the rest of this guide applies only alongside that. If the fake is part of a threat, a blackmail attempt or a demand for money, treat it as a crime in progress: keep the messages, do not pay and contact the police before you contact a platform.

## Make safety and support the first step, and avoid the common mistakes

Your first job is not removal: it is making sure you are safe and not alone. A sexual deepfake in particular is a form of abuse, and being targeted is not a sign of anything you did. If you feel unsafe, if there are threats or if someone knows where you live, contact the police or a local support service before you spend an hour on reports. Many countries have helplines for image-based abuse; the ones available where you live were not verified for this guide, so search for your own country's service.

Then avoid the mistakes that make it worse.

- **Do not confront the poster.** A reply can prompt deletion of the evidence, a copy elsewhere or more attention.
- **Do not pay.** Paying a blackmailer rarely ends the demand, and paying a "removal service" that approached you first is a common second scam.
- **Do not repost the fake to say it is fake.** Even a denial that includes the file spreads it and can break platform rules against hosting it. A short statement in words, without the file, does the job.
- **Do not send the file to friends to ask if it looks real.** Describe it in words to the one or two people you trust, and let them decide how much they need to see.
- **Do not delete your own accounts in a panic.** The accounts hold your evidence and your ability to report as the verified owner of your own identity.

<aside class="callout">
<span class="callout-label">Warning</span>
<h4>If the person shown is under 18, do not copy, save or forward the file</h4>
<p>Screenshots and downloads of sexual content showing a minor can raise serious legal problems in many places, even when you are the victim. Note the page address and report it to the police and a child-protection service in your country, not only to the platform.</p>
</aside>

## Preserve evidence without copying or spreading the file

Reported content disappears, and later requests, police reports and court papers depend on proof of what was posted, where and when. Keep a record of the page, not a collection of copies.

<figure class="fig">
<figcaption><b>Record the deepfake in six steps</b></figcaption>
<ol class="fig-steps">
<li><b>Open a private log</b>Create one document that only you can read. Put in the date and time you found the fake and how you found it, in words.</li>
<li><b>Copy the page address</b>For every place the fake appears, record the full address, the account name that posted it and the page title. The address is the single most useful item in any report.</li>
<li><b>Capture the surrounding text</b>Take a screenshot of the post's caption, the comments and the account page, cropped so that a sexual fake is not shown. For sexual content, note what it depicts in one plain sentence instead of capturing it.</li>
<li><b>Note the reach</b>Write down view counts, shares, and the number of copies you know of. This supports urgency in a report and later any claim of harm.</li>
<li><b>Record what you did</b>For each report, write down the platform, the time, the category chosen, the reference number and any reply. Platforms often close a report without detail, and the log is how you escalate.</li>
<li><b>Store the log safely</b>Keep it off shared devices and out of a cloud folder that others can open. Give it to a lawyer or the police if you decide to act.</li>
</ol>
</figure>

## Report the deepfake to every platform that hosts it

A report to the platform is the fastest way to remove the copy on that platform, and it is the step that most often decides how long the fake stays public. Most large platforms have policies against intimate imagery shared without consent, against impersonation, and against manipulated or synthetic media in some forms. The exact wording differs by platform and changes often, and those policy pages could not be opened for this guide, so read the live policy and pick the category that fits what the fake does.

### What to put in the report

A good report is short and specific. It states that you are the person shown, that the content is synthetic or manipulated, that you did not consent to it, and the one policy category it breaks. It gives the exact page address and, where the form allows, asks that copies be removed as well. It does not argue the case at length or include the file.

- **Use the category for non-consensual intimate imagery** for sexual fakes, and add that it is synthetic. Reviewers often route these differently from ordinary harassment reports.
- **Use the impersonation category** when the fake pretends to be you, and use the scam category when it is used to take money.
- **Use the manipulated media category** where one exists and the fake puts false words or acts in your mouth.
- **Use a legal request form only where one exists** for your situation, because legal forms often route to a different team with a different standard.

### Go beyond the first platform

Fakes rarely stay in one place. After the first report, work down the chain: the sites that reposted it, the hosting provider of a standalone site, and the registrar for a site built to publish this kind of content. Hosting providers and registrars usually have an abuse contact and acceptable-use terms, and many act on a clear report of non-consensual content, although deliberately rogue operators often do not respond.

Keep a note of every reply. A refusal is useful: it tells you the platform's reasoning, it allows an appeal and it is the evidence a lawyer needs to show you tried the ordinary route first.

### Appeal a refusal once

If a platform rejects a report, resubmit once with the correct category and a clearer statement, or use its appeal option. Repeated identical reports do not change a reviewer's view and can flag your account. Move to the next route instead.

## Use hash matching so known copies are stopped from returning

Hash matching is a way to stop copies from being uploaded again, and for sexual fakes it is often the most efficient step after the first reports. <a href="https://stopncii.org/how-it-works/">StopNCII.org</a> describes how it works: you select the intimate images or videos on your device, and the tool creates a hash, a digital fingerprint of each file. Only the hash is sent from your device, not the image or video itself, so the content stays with you. Participating platforms compare the hash against uploads and remove matches that break their intimate image abuse policy, and the service says it periodically keeps looking for matches. Keep the case number and PIN it gives you, because the page says a lost PIN is not recoverable.

Read the limits as carefully as the promise. The same page says StopNCII.org cannot remove images from the whole internet, only from the participating platforms on its partners page, so check that list for the places where your fake appears. The how-it-works page does not mention synthetic images, but the service's FAQ does: it says manipulated images and deepfakes are included in its definition of intimate image abuse. The FAQ sets conditions. The image must show you, you must have access to it, it must be nude or sexual, and you must be over 18 in it. For an image of someone under 18, the FAQ points to a separate service, TakeItDown, and you cannot submit on behalf of another person. Because the tool works from a file on your device, use it only if you already hold the file and are comfortable creating the hash yourself.

## Ask search engines to stop listing the fake

Search removal does not delete a deepfake, but it removes the thing that makes it easy to find under your name. A search engine can act only on its own results, so the page stays live on the site until the host or a court acts. Treat search requests as a parallel track to platform reports, not a replacement.

Google has a help page titled <a href="https://support.google.com/websearch/answer/9116649">Remove explicit non-consensual fake imagery from Google</a>. It covers fake sexual or nude imagery in which you are identifiable and that was distributed without your consent. You can file as the person shown or through a representative, you give the URLs of the pages (only those URLs are reviewed), and for sexual imagery Google says it does its best to find and remove duplicates, which you can opt out of on the form. After approval the URL no longer appears in Search, in full or only for searches with your name, and the content can still exist on the host site, which you must contact separately. A fake that is not sexual falls outside that page, so read the live policies for other kinds. Other search engines have their own request forms, so repeat the request wherever the fake is listed.

## Use copyright notices only where they genuinely apply

A copyright notice is a formal request that has a real basis only when you own the copyright in the material the fake was built from, such as a photograph you took, a video you filmed or a recording you made. In the United States this process is commonly called a DMCA notice. If someone else took the original photo or video, you usually do not own it and the notice would not fit. Because a notice is a formal statement, send one only if the facts are true, and ask a lawyer when unsure. The US Copyright Office page on <a href="https://www.copyright.gov/512/">section 512 of the DMCA</a> says a valid notice needs a signature, identification of your work and of the infringing material, your contact details, a good faith statement and a statement under penalty of perjury that you are authorized to act. Rules outside the United States could not be read for this guide.

## Take the legal route when platforms refuse, fail or cannot reach the fake

Legal routes are slower and more costly than reports, and they are the only route for some situations: a site that ignores abuse reports, a poster you can identify, a fake used to extort or defraud, or a fake that is damaging your work. Which law applies depends on where you live, where the poster is and where the site is hosted, and some places have added specific rules on deepfakes. The rules in your jurisdiction were not verified for this guide, so the points below describe the types of route, not the answer for you.

### A federal removal duty in the United States

The United States has a federal law called the TAKE IT DOWN Act, which concerns non-consensual intimate images, including digitally forged ones. The <a href="https://consumer.ftc.gov/node/331881">FTC's consumer page</a> says it applies to covered platforms such as social media, messaging and image or video sharing sites, covers real images, digitally altered images and AI deepfakes, and requires removal of the content and known identical copies within 48 hours of your request. If a platform misses that or has no working process, the FTC page says you can report it at TakeItDown.ftc.gov, and you can report for someone else with their permission. The FTC page does not say what a request must contain, and the text of the law could not be read, so check the platform's own removal form or ask a lawyer to send the notice.

### Criminal reports

Sharing sexual images of a person without consent, extortion, harassment and fraud can be offenses in many places, and in some of them it does not matter whether software made the image. A police report creates an official record that platforms and lawyers often ask about. Bring your log and ask for a crime reference number.

### Civil claims and court orders

Civil claims can include misuse of your likeness, invasion of privacy, harassment or defamation where the fake is false and damages your reputation, depending on the country and state. A court can order a poster or, in some systems, a platform to remove content, and it can order a search engine to de-list a result where a local law allows. This is where a specialist lawyer earns their cost: the choice of claim, the forum and whether to name an unknown poster all change the outcome. Ask any lawyer about cost and risk before you start, including the risk that a public case draws more attention to the fake.

## What to expect: a realistic timeline, and what to do when a route fails

These timings are our experience, not a statistic or a promise. They assume a fake on a mainstream platform; a rogue site or a hostile poster can double them.

<div class="tablewrap"><table><thead><tr><th>Stage</th><th>Usually</th><th>What you are doing</th></tr></thead><tbody>
<tr><td><strong>First hours</strong></td><td>Same day</td><td>Safety check, log, first platform reports, bank or police if money or threats are involved</td></tr>
<tr><td><strong>First days</strong></td><td>One to three days</td><td>Reports to every other host, search requests, hash matching where it fits, an appeal if refused</td></tr>
<tr><td><strong>First weeks</strong></td><td>One to four weeks</td><td>Follow up reference numbers, report new copies, contact hosts and registrars, speak to a lawyer if refused</td></tr>
<tr><td><strong>Months</strong></td><td>Several months</td><td>Legal letters or court steps where needed, monitoring for reposts</td></tr>
</tbody></table></div>

### An invented scenario

This scenario is invented and uses generic parties. Imagine a self-employed trainer, "Alex", who finds a fake video in which their face and voice are used to promote a supplement in an online advert. A client sends the link. Alex checks safety first, which is simple here because there is no threat. Alex logs the advert's page address, the advertiser's account name and the number of views, and does not download the video. In the first hour Alex reports the advert to the ad platform as impersonation and deceptive advertising, then reports the same video through the ordinary report on the video site. Alex posts a short text note on their own account saying the advert is not theirs and not to buy, without reposting the video.

By day two the ad platform has removed the advert, and the video site has not responded. Alex files one appeal with a clearer description and, as the advertiser's account is still running other adverts, reports those too. Alex also tells their bank and checks that no one has used the voice to approach clients. After two weeks the video site has still not acted, and Alex asks a lawyer to write to the site's host. The lawyer's letter, not the third report, gets a response. In this invented case the cost was time, a small legal fee and some weeks of watching, and the result was removal from the main hosts, not from the whole internet. That is a realistic outcome.

### When a route fails

Each route has a known failure point. Knowing it saves time.

<div class="tablewrap"><table><thead><tr><th>Route</th><th>Where it usually fails</th><th>What to do next</th></tr></thead><tbody>
<tr><td><strong>Platform report</strong></td><td>Wrong category, no response, or a site that ignores abuse</td><td>Resubmit once with the right category, then go to the host</td></tr>
<tr><td><strong>Hash matching</strong></td><td>Site is not a partner, or the image does not meet the eligibility conditions</td><td>Check the partner list and the FAQ, use direct reports</td></tr>
<tr><td><strong>Search request</strong></td><td>Content not covered by the policy, or page is lawful and accurate</td><td>Ask the host or use a legal order</td></tr>
<tr><td><strong>Copyright notice</strong></td><td>You do not own the source material</td><td>Use the platform's consent-based policy instead</td></tr>
<tr><td><strong>Legal letter or claim</strong></td><td>Poster unknown, site abroad, or cost out of proportion</td><td>Ask a lawyer about a court order or a host-level request</td></tr>
</tbody></table></div>

<aside class="callout">
<span class="callout-label">Limit</span>
<h4>No route removes a fake from the whole internet</h4>
<p>Copies can sit on private channels, small sites and devices that no request reaches. The realistic goal is to remove it from the places where people find it, stop re-uploads on cooperating platforms and keep a record that supports a legal step.</p>
</aside>

## Get help when the fake is moving faster than you can report it

Ask for help if the fake is sexual, spreading across several platforms or used against your work. A specialist can run the reports, keep the log and escalate to hosts and lawyers while you step back. Reputation Partners takes on this kind of file, and it begins with a private audit of where the fake appears and which route fits each copy.

<div class="box hot"><b>If this is happening to you now</b>
<p>Start with your log and the first reports, and contact the police if there is a threat or the person shown is under 18. If you want a team to take over the reports and the escalation, the leaked and intimate content removal page explains how we handle these files, or you can ask for a private audit there. See <a href="/leak-removal">leak and intimate content removal</a>.</p></div>

## Questions people ask

### Can a deepfake be removed from the internet completely?

Not with one request. A deepfake can be removed from the platforms and search results that act on your reports, and known copies can be blocked from re-uploading on cooperating platforms. Copies on private channels, rogue sites or personal devices may remain. The realistic aim is to remove it from public view, keep a record and stop it returning where you can.

### Is it illegal to make a deepfake of someone?

It depends on the country and state and on what the fake shows and is used for. Sexual fakes made without consent, fakes used for fraud or extortion and fakes used for harassment can be offenses in many places, and some places have added deepfake-specific rules. The rules where you live were not verified for this guide, so ask a lawyer or your police force.

### Should I save the deepfake as evidence?

Usually no. Record the page address, the account name, the date and what the caption says, and describe the content in words. Saving a sexual fake adds a copy, and saving content that shows a minor can create legal problems of its own. If a lawyer or the police need the file, they will tell you how to hand it over safely.

### Will reporting a deepfake make it spread further?

A report goes to the platform's reviewers and is not public. The risk of spreading comes from reposting the fake yourself, replying to the poster in public or sending the file to friends. Report in private, keep your statement in words and do not engage. If a legal case could draw attention, a lawyer can weigh that before you file.

### Can I sue the person who made it?

Sometimes. A claim needs a poster you can identify, a legal basis in your country or state and a case worth its cost. Civil claims, criminal reports and court orders each work differently. A specialist lawyer can tell you whether to start with a letter, a claim or a request to a host, and what each is likely to cost and risk.

### What if the site hosting it ignores me?

Escalate in steps. Report to the site's hosting provider and the domain registrar through their abuse contacts, then ask a lawyer to send a formal letter or seek a court order. Ask search engines to stop listing the pages while you do this. Some rogue sites never respond, and in that case search removal and monitoring are the realistic protections.