Back to blog
Piracy & Leaks8 min readDMCA Master TeamLast reviewed

Deepfakes and AI-Generated Impersonation: Removal Options in 2026

AI-generated videos, voices and images of real people are cheap to make. This guide covers the removal routes that work in 2026: platform policies, NCII rules, impersonation reports, copyright and new laws.

Written by the DMCA Master team, who prepare and file takedown reports for clients. General information, not legal advice.

A finance influencer wakes to messages from followers asking why she is promoting a crypto scheme. She is not. A video on Instagram and TikTok shows her, in her own voice and studio, urging viewers to invest through a link. Every word was generated. By the time she sees it, the video has been reposted by a dozen accounts and some of her followers have lost money.

A university lecturer finds that someone has used his face in explicit AI-generated images posted to a forum. A small business owner discovers a cloned voice of her founder in a phone scam targeting her customers. Each of these is a different kind of harm, and in 2026 each has a removal route that works, provided you pick the right one. This guide covers deepfake removal and AI impersonation across the platforms and legal frameworks that matter now.

Three kinds of deepfake harm, three sets of rules

Platforms and laws treat synthetic media differently depending on what it depicts. Identify which category you are in before filing.

Category Examples Fastest route
Sexual or intimate synthetic content AI-generated nude or sexual images or videos of a real person NCII (non-consensual intimate imagery) policies and laws; Google's explicit synthetic content removal
Deceptive impersonation Fake endorsements, scam ads, cloned voices, fabricated statements Platform impersonation and scam policies; synthetic media rules; trademark where a brand is used
Non-sexual, non-deceptive synthetic likeness Parody, satire, "what if" videos, face-swapped memes Limited; privacy complaints on some platforms; right of publicity for commercial use

The first two categories are highly removable in 2026. The third is largely protected expression in most jurisdictions unless it is used commercially or crosses into defamation.

Sexual synthetic content: the fastest route

Platforms now treat AI-generated intimate imagery of a real person exactly like real non-consensual intimate imagery. Use the NCII report option on each platform rather than a general report; it reaches safety teams and is prioritized.

  • Meta, X, TikTok, Reddit, Snapchat, Discord: each has a report category for intimate imagery shared without consent, and each applies it to synthetic content.
  • Google Search: has a dedicated removal request for explicit fake or AI-generated imagery depicting you, separate from the standard personal-content and copyright forms. Google removes the URLs from search results and has stated that it also works to filter similar results for your name.
  • Bing: equivalent form; covers Yahoo and DuckDuckGo results too.
  • Independent sites and forums: notify the host (via Cloudflare's abuse form if proxied), citing both the host's acceptable-use policy and the applicable law.
  • StopNCII.org: lets adults hash real intimate images to block re-uploads on participating platforms; it is less useful for synthetic content, where each generated image differs, but worth using where a specific image is being spread.

The legal backing is now substantial. In the US, the TAKE IT DOWN Act (2025) criminalizes publishing non-consensual intimate imagery including AI-generated depictions and requires covered platforms to remove reported content within 48 hours. Many US states have their own deepfake intimate imagery laws. In the UK, creating or sharing sexually explicit deepfakes is a criminal offence and the Online Safety Act places duties on platforms to remove them. The EU's 2024 directive on violence against women covers non-consensual sharing of intimate material, including manipulated material, as it is implemented across member states. Australia's eSafety Commissioner has formal removal powers that extend to synthetic intimate imagery.

Cite the relevant law in your reports to hosts. Our leaked private content guide covers the NCII process in more depth, and our adult content protection service handles these cases confidentially.

Deceptive impersonation: scams, fake endorsements, cloned voices

Here the strongest routes are impersonation and scam policies rather than copyright, because the content is new, not copied.

Platform reports

  • Meta (Facebook, Instagram): report the post or ad as a scam or as impersonating you; for ads, use the ad's report option and select misleading or scam. Meta also has a process for public figures and businesses to report celeb-bait scam ads. Report the accounts posting the video, not just the video.
  • TikTok: report for impersonation and for "synthetic and manipulated media" misleading viewers; TikTok's policy prohibits synthetic media depicting real private figures and requires labelling for realistic AI content.
  • YouTube: the privacy complaint process now covers AI-generated content that simulates an identifiable person's face or voice. File it from the video's report menu, choosing privacy. This is separate from copyright and does not need you to own anything.
  • X: report under synthetic and manipulated media and impersonation policies.
  • LinkedIn: report fake profiles and posts using your identity through the profile's report option; LinkedIn acts quickly on verified executive impersonation.

Our impersonation protection service and fake profile removal guide cover the account-level actions.

Brand and trademark angles

If the deepfake uses your company name, logo or product, file a trademark report with the platform alongside the impersonation report. Trademark reports are processed by IP teams and succeed even where an impersonation report is borderline.

Copyright, where it applies

Deepfakes are often built on your real footage or photos. If the fake incorporates your copyrighted video, images or audio (a studio shot, a podcast clip), a DMCA notice for that underlying material is a valid and fast route. State clearly which original work has been reproduced. Do not use copyright for purely generated content you did not create; a knowingly false notice carries liability under Section 512(f).

Scam infrastructure

Scam deepfakes point somewhere: a website, a payment page, a phone number. Report the website to its host and registrar as a phishing or fraud site, report the domain to Google Safe Browsing, and report the payment processor. Removing the destination protects your audience even before the videos come down.

Search

De-index any pages hosting or promoting the fake through Google's and Bing's removal processes: copyright where your material is used, and personal content or policy grounds where it is deceptive content about you. Our search engine de-indexing service runs these.

Non-sexual, non-deceptive synthetic likeness

A satirical video that is clearly labelled, a parody using your voice, a meme with your face: in the US these are usually protected speech, and in the UK and EU they often fall under parody and free expression exceptions. Platforms will typically not remove them unless they are deceptive, harassing or sexual.

Exceptions worth pursuing:

  • Commercial use. Using your likeness or voice to sell something without consent violates right-of-publicity laws in many US states (Tennessee's ELVIS Act, for example, explicitly covers AI voice cloning) and passing-off or personality-rights principles elsewhere. Platforms act on these as impersonation or misleading-advertising violations.
  • Defamation. A fake video depicting you saying or doing something that damages your reputation and that viewers believe is real may be defamatory. That is a legal claim and a basis for a platform report, but it requires care; see our brand defamation removal guide.
  • Labelling requirements. The EU AI Act requires deepfakes to be disclosed as artificially generated; several platforms require AI labels. Unlabelled realistic synthetic content can be reported for the missing label even where it would not be removed outright.

A sequencing checklist

  1. Capture evidence: screenshots with URLs and dates, the account names, and a copy of the video if it is safe to store (not for intimate content — screenshot the page with the image cropped out).
  2. Identify the category: sexual, deceptive or expressive.
  3. File the category-specific report on every platform where it appears, plus accounts that are reposting.
  4. File supporting reports in parallel: trademark (brand used), copyright (your footage used), scam (destination site).
  5. De-index pages from Google and Bing under the appropriate policy.
  6. Notify hosts of independent sites, citing applicable law.
  7. Warn your audience briefly and factually if the content is a scam targeting them; do not amplify intimate content.
  8. Re-check all URLs at 48 hours and one week; report reposts as repeats.
  9. Where a perpetrator is identifiable, assess legal action with counsel.

Timelines and expectations

In our experience, synthetic intimate content is removed from mainstream platforms within a day or two and from search within three; scam deepfakes are removed from Meta and TikTok within one to three days but are reposted aggressively while the scam runs, so monitoring during the campaign is essential; expressive content is rarely removed. DMCA Master handles deepfake campaigns at $100 per successfully removed URL for single cases (no removal, no fee) or $10 per hour for ongoing protection with a dedicated specialist working 40 hours per week, billed weekly ($400/week), cancel any time.

Frequently asked questions

Do I need to prove the video is fake? For NCII and impersonation reports, a sworn statement from the person depicted that they did not consent and did not make the content is usually enough. Platforms may apply their own detection. Keep any evidence that helps (you were elsewhere, the original footage).

Can I use copyright to remove a deepfake? Only for the parts you own: your original footage, photos or recordings used as source material. Not for purely generated content.

What if the deepfake is on a site that ignores reports? De-index it, notify the host and upstream provider, cite the applicable criminal law, and report any scam infrastructure. The hidden copy matters far less once it cannot be found.

Is a parody deepfake of me illegal? Usually not, unless it is sexual, commercial, defamatory or deceptive. Labelling rules may apply.

Who do I report voice cloning scams to? The platform carrying the audio or ad, the phone carrier or VoIP provider where calls are involved, your national fraud reporting body, and the payment processor. Where your brand is used, add a trademark report.


This article is general information, not legal advice. If a deepfake or AI impersonation of you or your brand is circulating, request a free analysis and a specialist will map every copy and start removals within 24 hours.

Share: X · LinkedIn · Facebook

Related articles