Free· Confidential· Step by step · You haven't done anything wrong.
removemydeepfake
A magnifier and notebook on a warm desk, suggesting careful examination and record-keeping

How to Tell If Something Is a Deepfake, and Prove It

Reviewed against primary sources Updated 2026-06-06 · The Remove My Deepfake Project

How to spot a deepfake, the detector tools that can help, and the critical step most people skip: preserving evidence (URLs, screenshots, archives) before you report, plus how to find other copies with reverse image search.

To tell if something is a deepfake, look for mismatched lighting, blurring where the face meets hair or neck, odd eyes, teeth, or hands, and, in video, unnatural blinking or off lip-sync. Detector tools can help but are not certain. Most importantly, before you report anything, preserve the evidence: URLs, dated screenshots, and archived copies.

Do this first

Content often vanishes the moment you report it. Preserve evidence before reporting, so you keep proof for platforms, police, or a lawyer. The steps are below.

How to spot a deepfake

No single sign is proof, but several together are a strong signal. In images, look for:

  • Lighting or skin tone that does not match between the face and the rest of the body.
  • Blurring, smearing, or warping where the face meets hair, ears, neck, or glasses.
  • Irregular eyes, teeth, or jewellery, and asymmetric or melted-looking details.
  • Backgrounds that bend or repeat oddly near the edges of the figure.
  • Hands and fingers that look wrong, a common giveaway in AI images.

In video, also watch for unnatural or rare blinking, lip movements that do not quite match the audio, flickering at the edge of the face, and a face that seems to “float” slightly against the head when the person moves.

Detector tools can help, but are not proof

Several services claim to detect AI-generated or manipulated media. They can be a useful second opinion, but accuracy varies and the technology is a moving target on both sides. Treat a detector result as supporting evidence, not a verdict, and never rely on it alone when escalating.

Preserve the evidence (step by step)

This protects you for every route that follows, platform reports, Google removal, police, or a lawyer.

  • Copy the exact URL of every place you find it.
  • Take dated screenshots that clearly show the URL, the date, and the account that posted it.
  • Archive each page using the Wayback Machine or archive.today, so a copy exists even if the original is deleted.
  • Record the account details (username, profile link, display name) and a simple log of dates and times.
  • Keep any source photo the fake was likely built from; it can help prove manipulation.

Store all of this somewhere safe and private. You may never need it, but if you do, you will be very glad you saved it.

Find the other copies

Deepfakes often appear in more than one place. To find copies you have not seen:

  • Reverse image search a screenshot or frame using Google Lens or another reverse-image tool.
  • Search your name together with related terms the way others might.
  • Set up alerts in Google’s “Results about you” so you are notified when new results mentioning you appear, see removing a deepfake from Google.

What not to do

  • Do not contact or threaten the person who posted it. It can escalate the situation and complicate any case.
  • Do not re-share or repost the content, even to warn others; it can spread it further and may flag your account.
  • Do not delete your evidence after the content comes down.

Now take it down

With evidence saved, move to removal: register intimate images with StopNCII or Take It Down, report it on the platform, and remove it from Google. If you are not sure of the order, the action-plan tool on our home page builds a sequence for your exact situation.

Frequently asked questions

How can I tell if an image is a deepfake?
Look for tell-tale signs: mismatched lighting or skin tone, blurring or warping where the face meets hair, ears, or neck, irregular teeth or eyes, inconsistent backgrounds, and odd hands. With video, watch for unnatural blinking, lip-sync that is slightly off, and flickering at the edges of the face. Detector tools can assist but are not foolproof.
How do I prove something is a deepfake?
Preserve strong evidence: save the original URL, take dated screenshots showing the URL and posting account, archive the page (Wayback Machine or archive.today), and keep any original photo the fake was likely built from. Detection-tool reports and an expert opinion can support your case if you escalate legally.
What evidence should I save before reporting a deepfake?
Save the exact URLs, dated screenshots that capture the URL and the account, an archived copy of each page, the usernames and any profile details of who posted it, and a simple log of dates and times. Do this before reporting, because content often disappears once you report it.
Are deepfake detector tools reliable?
They can help but should not be your only proof. Detection accuracy varies and the technology keeps changing on both sides. Use detector results as supporting evidence alongside preserved URLs, screenshots, and archives, not as a guarantee.
How do I find other copies of a deepfake online?
Use reverse image search (such as Google Lens or other reverse-image tools) on a frame or screenshot, and search your name with related terms. Set up Google’s “Results about you” alerts so you are notified when new results appear.

Not sure where to start?

Answer a few quick questions and get a step-by-step plan built for your situation. Free, and nothing you type leaves your device.

Start your Takedown Plan →