How to Remove a Deepfake of Yourself From the Internet

Last updated: May 2026

The first thing most people do when a deepfake of them surfaces is panic. The second thing they do is start clicking. They click report on the post, they click contact us on the host, they Google “how to remove a deepfake of me,” and they end up on a tangle of half-answered help pages, expired removal forms, and Reddit threads from 2022. The clock keeps running while the file keeps spreading.

This post is the playbook we wish every client had before they called us. It walks through what counts as a deepfake under current US law, which removal channels actually work in 2026, what to do in the first 24 hours, and where to escalate when the platform ignores you. It is not legal advice. It is the same operational sequence we run for clients whose face, voice, or likeness has been used without permission in synthetic media.

By the end you will know exactly where to send a notice, what evidence to attach, and how to make sure the file does not quietly reappear under a new URL next week.

What Counts as a Deepfake in 2026

A deepfake is synthetic media generated or altered by machine learning that depicts a real person doing or saying something they did not actually do. The term originally referred to face-swap pornography in 2017, but the category has expanded to include voice clones, full-body generative video, hybrid edits, and AI-cloned avatars used in scams and harassment. The Brookings Institution’s tracker on deepfake regulation catalogs how policy language has caught up to the technology over the past three years.

Two categories matter for removal. The first is non-consensual intimate imagery, including AI-generated sexual content. The second is everything else, which covers fraudulent endorsements, fake “confessions,” impersonation videos used to defame or harass, and voice clones used for extortion or financial fraud.

The removal path is different for each, and a misclassified request gets denied or routed to a slow queue. Identify which bucket your situation falls into before you file anything.

The Law Has Finally Caught Up

For most of the last decade, deepfake removal was governed by a patchwork of state laws, narrow federal statutes that never quite applied, and platform terms of service that were inconsistently enforced. That changed in 2025.

The TAKE IT DOWN Act, signed into federal law in May 2025, criminalizes the publication of non-consensual intimate visual depictions, including AI-generated ones, and requires covered platforms to remove reported content within 48 hours of a valid notice. The Federal Trade Commission’s statement on the law and the bill summary published by Senator Cruz’s office lay out the obligations on hosts. For sexual deepfakes of adults and minors, the 48-hour removal clock is now federal.

Older laws still matter for the rest of the deepfake universe. The Copyright Act and its DMCA takedown framework handles cases where source footage of yours was scraped to train or seed the synthetic media. State right-of-publicity statutes cover unauthorized commercial use of your face, voice, or name. State criminal codes increasingly include deepfake-specific provisions, with California, Texas, Virginia, New York, and Illinois among the most aggressive. The pending NO FAKES Act would extend right-of-publicity-style federal protection to voice and likeness against AI replication.

The EU AI Act’s transparency obligations, which began to phase in during 2025, require deepfake content to be labeled as artificially generated and give EU residents stronger removal levers when content is hosted or accessed there. The UK Online Safety Act and Australia’s eSafety Commissioner framework add further teeth in those jurisdictions. If the content is hosted, mirrored, or accessible outside the US, the international frameworks are worth knowing because they sometimes move faster than US channels.

The First 24 Hours

When a deepfake of you surfaces, the first day matters more than the next thirty. Files mirror. Screenshots get rehosted. Search indexes update. The shape of the problem on day two is rarely the same shape it had on day one.

Five things should happen in that window.

Preserve the evidence. Before you do anything else, capture full-page screenshots of every URL hosting the file, the URL bar visible. Save the file itself (yes, the deepfake) to a private drive that nobody but you and your counsel can access. Note the exact timestamps you observed each URL. The Reporters Committee for Freedom of the Press and the Electronic Frontier Foundation both publish evidence-preservation guidance that applies cleanly here. If you skip this step and the platform takes the content down before you file anything else, you have lost your record of what was published, where, and when.

Identify every URL. Run a reverse image search and a reverse video search across the major engines. Google’s reverse image search, Bing’s visual search, TinEye, and Yandex Images catch different mirrors. Save the URL list. Most clients are only aware of one or two URLs when they call us, and the actual mirror count is closer to five or ten.

Do not engage with the poster. Replies, DMs, screenshots of your reaction, all of it feeds the engagement signals that keep the file ranking. Resist the urge to threaten, plead, or correct the record in the comments. Every reply makes the post stronger.

File the platform reports. For each URL, file the appropriate report through the host’s own form. Specific links are below.

Loop in counsel and, if criminal, law enforcement. For sexual deepfakes, law-enforcement reporting is part of the standard sequence even when removal is the priority. The National Center for Missing and Exploited Children’s CyberTipline handles minor cases. For adults, the local field office of the FBI and state attorneys general with deepfake or revenge-porn units are the standard entry points.

Platform Removal Forms, By Platform

The single most important practical fact about deepfake removal in 2026 is that almost every major platform now publishes a dedicated form for synthetic and non-consensual intimate imagery. Filing through the right form drops your case into a faster queue than filing a generic “report this content” complaint.

Google Search. Use Google’s removal request for explicit or intimate personal images. The same form covers AI-generated explicit imagery as of 2024. For non-explicit deepfakes that are defamatory or impersonating, use Google’s “remove personal information” tool and the legal removal request form. Google does not host the content. It de-indexes it from the search results, which is usually what you want because that is what most people actually see.

Bing. Use Microsoft’s report a concern form and the Bing content removal tool. Microsoft has been one of the more responsive engines on synthetic NCII since 2024.

Meta (Facebook, Instagram, Threads). Use Meta’s dedicated intimate imagery removal flow and, for AI-generated intimate content, the partnered StopNCII.org hashing tool, which lets you submit a hash of the image without sending the image itself. Meta’s reporting policies for synthetic and manipulated media cover the non-intimate deepfake cases.

X (Twitter). Use the synthetic and manipulated media report flow and the non-consensual nudity policy. X has been inconsistent on response time historically. Filing under the correct policy materially improves the odds.

TikTok. TikTok’s synthetic and manipulated media policy requires labeling and prohibits non-consensual sexual deepfakes outright. Report through the in-app flow and, if the in-app report fails, the TikTok legal removal request form.

YouTube. Use the YouTube privacy complaint process, which now explicitly includes synthetic content, and the defamation removal form for false statements of fact.

Reddit. Use the involuntary pornography report form and the impersonation report flow. Reddit also removes mods of subreddits that repeatedly host this content, so subreddit-level escalation through contact@reddit.com sometimes helps when post-level reports stall.

Adult sites. Almost every major adult tube now requires uploader ID verification and has an expedited NCII removal queue. Use the dedicated content removal forms on each property. The Cyber Civil Rights Initiative’s image abuse helpline maintains an up-to-date guide to each platform’s current form, which is more reliable than us listing them by URL today and watching the URLs rot in six months.

Hashing-based removal for NCII. StopNCII.org, run by the Revenge Porn Helpline and partnered with Meta, TikTok, Bumble, Reddit, Snap, OnlyFans, Pornhub, and others, lets adults hash an intimate image or video locally on their device and share only the hash. Partner platforms scan for the hash and block matching uploads. For minors, TakeItDown.NCMEC.org does the same with the same major platform partnerships. Hashing-based removal is the most effective tool we have for stopping re-uploads at scale, and it is underused because most people do not know it exists.

When Platform Reports Stall

Not every report works the first time. Three escalation paths are worth knowing.

The legal-route notice. A demand letter from counsel, citing the specific platform policy, the relevant state or federal law, and the URLs in question, is treated differently by Trust and Safety than a user report. It often unlocks a manual review queue and shortens the response window from days to hours. The American Bar Association’s guide to NCII civil remedies is a useful starting point for understanding what the letter needs to contain.

The DMCA route, where it applies. If the deepfake incorporates copyrighted material you own, such as a photo or video you originally posted, a properly drafted DMCA notice gets the content removed under federal law regardless of the platform’s deepfake policy. The notice has to follow the specific statutory form. Sloppy notices get rejected.

The de-indexing route. Even when a host refuses to remove a file, search engines will usually de-index the URL under the explicit-imagery removal policies above. De-indexing means the page still exists somewhere but does not surface in normal Google or Bing results, which is the layer most counterparties (employers, dates, journalists) actually see. We cover the broader mechanics of suppressing negative search results elsewhere on the site.

When all three escalation paths fail, civil litigation is the last lever. State right-of-publicity, defamation, and intentional infliction of emotional distress claims, paired with the federal TAKE IT DOWN Act for sexual cases, give plaintiffs working causes of action in every state. Subpoenas to the host or upstream provider can identify the original uploader, which is often necessary for actual damages and for stopping the cycle at the source.

The AI Search Layer

Removing the file from the open web is half the job in 2026. The other half is making sure the AI search layer is not still surfacing it.

ChatGPT search, Perplexity, Gemini, and Microsoft Copilot generate their answers by retrieving from the open web and synthesizing. If a deepfake article still ranks in the underlying index, the AI answer about you may cite it long after the original post is taken down. We treat this as its own work stream. The standard sequence is to use each platform’s reporting flow for inaccurate or harmful content about a real person, push down the underlying citations using authoritative content you control, and monitor the AI answer over time to confirm it stabilizes. Our AI search reputation management practice is built around exactly this problem.

The Stanford 2024 AI Index Report and Pew’s 2024 research on AI chatbot usage both confirm that mainstream users are now starting their research about real people in AI chatbots before they touch Google. A clean search result that is undermined by a stale AI answer is not actually clean.

What the Data Says About Deepfake Prevalence

The scale problem is worse than most people assume.

The early Sensity AI (formerly Deeptrace) deepfake landscape report found that the overwhelming majority of deepfakes in circulation were non-consensual sexual content targeting women, and the share has not meaningfully improved since. The Cyber Civil Rights Initiative’s research on image-based abuse documents both the prevalence and the mental-health cost on victims. The WeProtect Global Threat Assessment and the Internet Watch Foundation’s 2024 reports show a sharp rise in AI-generated child sexual abuse material since 2023, which is why the federal and state legislative push has accelerated.

Outside the sexual category, the deepfake fraud trend is real and growing. The FBI’s Internet Crime Complaint Center 2024 annual report catalogs an increase in CEO and family-member voice-clone scams, and Verizon’s 2024 Data Breach Investigations Report flags synthetic media as an emerging social-engineering vector. Deepfake removal is no longer a niche reputational concern. It is increasingly a financial and security one.

What We Tell Clients About Outcome

Removal is achievable. Speed depends on which platform, which jurisdiction, and which removal channel the case fits into. Sexual deepfakes covered by the TAKE IT DOWN Act move fastest because the federal removal clock is 48 hours. DMCA removals usually clear within a week. Defamation and impersonation deepfakes vary widely. International mirrors and bulletin-board-style hosts are the slowest.

Stopping re-uploads is its own discipline. The hashing tools work, the de-indexing tools work, the legal route works. The combination is what holds.

We offer guarantees on outcomes across our crisis and content-removal work, weighted toward performance-based structures. We will tell you in the consult whether your situation is one where a guarantee makes sense, and what the realistic timeline looks like. We will also tell you when the right answer is to skip the work and go straight to litigation, because sometimes that is the right call.

Mistakes to Avoid

A short list of the things we see clients do that make removal harder.

Posting about the deepfake on your own channels. This signal-boosts the file, often catastrophically. If a public statement is required, it should come after the content is offline, not before.

Hiring a firm that promises to “scrub the internet.” There is no such product. Anyone selling one is selling marketing copy. Real removal is platform-by-platform, statute-by-statute, URL-by-URL.

Filing the wrong form. Generic content reports go to a slower queue than the dedicated NCII and synthetic-media forms. Read the platform policy first.

Filing without evidence. The fastest-moving Trust and Safety teams still require the URL list, screenshots, and proof of identity. Filing with a half-built record stalls the case.

Trying to handle the AI search layer with the same playbook as the open web. The AI answer is a different system with different removal levers. It needs its own work stream.

Ignoring the legal angle until later. Civil and criminal options are sometimes available immediately and sometimes time-barred if you wait. Loop in counsel early.

How DCM Approaches Deepfake Removal

Our content removal practice handles the platform-by-platform takedown sequence, including the dedicated NCII and synthetic-media channels and the legal-route notice when needed. Our individual privacy and personal information removal practice covers the underlying data brokers and aggregators that re-publish or syndicate the file. Our executive and individual crisis reputation management practice handles the wider response when the deepfake is part of an active reputational attack rather than an isolated post. And our AI search reputation management practice handles the chatbot layer once the open web is cleaned up.

We do not promise miracles, we explain exactly what is possible, and we put outcome-based guarantees on the work where the situation supports them.

If a Deepfake of You Is Live Right Now

The most useful thing you can do in the next hour is start the evidence record described above, then file the platform forms. If you want a second set of eyes on the sequence, or a faster lane through the platform Trust and Safety queues, book a free consultation with our team. We will tell you what is realistic, what the timeline looks like, and where the leverage actually sits in your specific situation. There is no charge for the conversation.

The internet did not invent this problem, but it scaled it. The removal infrastructure has finally scaled to match. The faster you start, the faster the file comes down.