How to Tell If a Dating Profile Photo Is AI-Generated (2026)

To tell if a dating profile photo is AI-generated, run three checks: search the public web for the same face, scan the image for the glitches AI still leaves behind, and watch how the person behaves. No single test is proof, but together they are decisive.

Modern tools can invent a photorealistic face that belongs to no real person. Because that face was never photographed and never posted anywhere, it has no online footprint, which is exactly what makes a reverse face search so useful: a real person turns up across public profiles, while a generated face turns up nowhere. Pair that with the visual tells below and a refusal to do live video, and you can usually make the call in minutes.

A few years ago, catfish had to steal a real person's photos. Now they can generate an original face on demand. Sites like thispersondoesnotexist use a model called StyleGAN to synthesize a brand-new human face every time you refresh, and general image tools can do the same. None of those people exist.

Scammers love these images for one reason: a stolen photo can be traced back to its real owner, but a generated photo cannot. There is no original to find, no other profile to expose the lie. That removes the single most effective check most people know about, the reverse image search.

The stakes are not small. The FTC reports that romance scams cost victims hundreds of millions of dollars a year, and AI-generated faces are now a standard ingredient in the fake profiles that start those scams. Knowing how to spot one is basic online-dating hygiene in 2026.

The single most useful check is a reverse face search, not a reverse image search. Here is the difference and why it matters for AI faces.

• Reverse image search (Google Lens, TinEye) looks for copies of that exact picture. It is great for catching a stolen photo that appears elsewhere, but an AI face made for a single profile has no copies, so you get zero results and learn nothing.
• Reverse face search looks for the same face across the public web, even in different photos, crops, and angles. Run it on a real person and you typically find their other public profiles. Run it on a generated face and you find nothing that holds together as one consistent identity.

So the footprint itself becomes the signal. A real, active adult almost always has some public trace. A face that appears absolutely nowhere, attached to a profile that is asking for your trust, is a warning worth taking seriously. It is not proof on its own, but combined with the tells below it is usually enough.

Want the mechanics of both approaches side by side? See reverse face search vs. reverse image search.

AI faces have gotten good, but they still slip on the small stuff. Open the photo as large as you can and check the edges and details, not the smile.

• Mismatched or mangled ears and earrings. One earring, two different earrings, or an ear that melts into the hair. Accessories are where generators fail most often.
• Off teeth. Too many teeth, uneven counts, or a smeared row that does not resolve into individual teeth.
• Asymmetric glasses. Frames that are different shapes on each side, or that dissolve into the temple or cheek.
• Warped backgrounds. Melted objects, doorways that bend, patterns that repeat wrong, or blurry "people" that are just blobs.
• Hair that defies physics. Stray strands that float over the background or fade into it with no root.
• Too-perfect skin. Airbrushed, pore-less complexion with oddly even lighting and no real-world blemishes.
• Garbled text. Any lettering in the shot, a sign, a logo, a shirt print, comes out as nonsense characters.
• Dead-center, identical eyes. The eyes sit in the exact same spot in every photo (more on this next).

You rarely need all eight. Two or three clear glitches, especially around ears, teeth, or the background, are a strong sign the image was generated.

Here is a fast tell that most people miss. Many generated face sets align every image so the eyes land in nearly the identical position on the canvas. It is an artifact of how the models are trained.

To use it: if the profile has more than one photo, or you can pull up the source generator, flip quickly between two images. On real photos the head tilts, the camera distance changes, and the eyes move around the frame. On a batch of AI faces, the eyes barely move, as if pinned in place, even though everything else changes. That eerie stillness in the eye line is a signature of aligned, generated portraits.

The flip side is just as telling: a genuine person usually has candid shots, group photos, and pictures taken at different distances and in different years. A profile with a single flawless headshot and nothing else is exactly what a generator produces.

"This photo is fake" can mean two very different things, and the check that exposes each one is different. It helps to know which you are dealing with.

• A stolen real photo (a catfish using someone else's pictures) does have an owner somewhere. Reverse image search and reverse face search will often surface the real person, whose name will not match the profile. That mismatch is the giveaway.
• An AI-generated face has no owner. Searches come back empty, so you lean on visual glitches, the eye-position trick, the single-photo pattern, and behavior instead.

That is why you run both searches. Empty results push you toward "generated," a real-name match pushes you toward "stolen," and either way the person you are talking to is not who they claim to be. If you suspect stolen images specifically, our guide on whether your online crush is real walks through confirming a real identity.

Image analysis gets you far, but scammers give themselves away in how they act. Weigh these alongside the photo:

• They will not do a live video call, or the "camera is always broken." A real match can wave on video in real time; a fake face cannot.
• The relationship moves fast. Intense affection within days, then a reason they cannot meet.
• Money enters the chat. An emergency, a customs fee, a can't-miss crypto tip. This is the point of the whole exercise.
• They push you off the app to a private channel quickly, away from the platform's safety tools.

Any one of these next to an unverifiable face should stop you cold. For a fuller pre-date routine, see how to check if someone you met online is real.

Move in this order, and protect yourself before you confront anyone.

• Do not send money, gift cards, or crypto, and do not share intimate photos or personal documents. This is the whole game for them.
• Capture evidence first. Screenshot the profile, the photos, and the chat, and save the profile URL. Accounts vanish the moment they are reported.
• Run the searches so you know whether you are dealing with a generated face or stolen photos, then stop responding.
• Report the profile to the dating app or platform so it can be removed for others.
• If money changed hands or fraud was attempted, report to the FTC and the FBI's IC3, and contact your bank or card issuer right away.

You do not need to prove the photo is AI beyond doubt. If the face has no footprint, shows the tells, and the person dodges video and steers toward money, you already have everything you need to walk away.