The hard part of verifying someone online isn't the tools β it's that you don't want it to be fake. By the time a profile has become a person you talk to every day, running a cold check on their photos can feel like a betrayal of the trust you've already extended. Scammers know this exactly, and the warmth is the leverage. The point of a routine is to take the decision out of your feelings and put it back on the evidence.
The steps below move from the fastest check to the one that matters most. You won't always need all seven β but the warmer the connection and the sooner it heads toward money, travel, or secrecy, the more worth it is to work through every one.
The 7-step check
Reverse-image-search their photos before anything else
Save the clearest profile picture, then upload it to Google Lens, TinEye, and Yandex Images β not just one of them. If the same face turns up under a different name, on a stock-photo site like Shutterstock or Getty, or attached to a model's Instagram, you have your answer. Scammers rarely shoot their own photos; they lift them from real people or stock libraries. This two-minute pass catches a large share of stolen-photo profiles on its own, which is why it's the first thing to do and not the last.
Crop to the face and run a face-specific search
A whole-image search matches the exact file; it misses a photo that's been cropped, mirrored, or color-shifted. So take a second pass: crop tightly to just the face and re-run it. Yandex is the standout here β it quietly built the strongest face matching of any free engine and routinely surfaces results Google and TinEye don't, including profiles on Eastern-European and Russian sites the Western engines barely index. If the person uses the same face across accounts with different names, a tight face crop in Yandex is what usually exposes it.
Ask for a live, unscripted video call β on the app you met on
A real person can hop on video; a scammer running a dozen conversations will stall with a broken camera, a bad connection, or a crisis that keeps coming up. Insist on a live call, not a pre-recorded clip, and keep it inside the platform where you met rather than moving to WhatsApp or Telegram. "We've talked for weeks but they've never once been free for video" is itself the finding.
Run a liveness test during the call
Deepfake video calls are no longer theoretical, so don't just confirm there's a face β confirm it's attached to a live body. Ask them to turn their head fully to the side, wave a hand slowly across their face, or stand up and change the lighting in the room. Real-time face-swap tools still glitch on exactly these movements: the mask smears at the edges, a hand passing over the face warps, profiles distort. It feels awkward to ask, but anyone genuine will laugh and do it. Resistance to a simple hand-wave is a louder signal than any answer they give.
Check whether the story holds up across sessions
Catfish profiles are built from borrowed parts, so the details drift. The city changes between chats, the job doesn't match the photos, the time zone they message from doesn't fit where they claim to be. Ask specific, checkable things β the name of their employer, a local landmark near them, what the weather's doing there right now β and see whether this week's answers match last week's. A real life is internally consistent without effort; a fabricated one needs maintenance, and the cracks show over time.
Know what reverse search can't catch
Here's the limit that matters in 2026: an AI-generated face has never existed anywhere, so it won't appear in any reverse image search. A clean result β no matches found β is not proof the person is real; it may just mean the photo was synthesized. This is why reverse search alone has stopped being enough. When the images come back blank, lean harder on the live-video and liveness steps above, because those test for a real human in real time, which a generated headshot can't fake on demand.
Apply the one rule that settles it: no money, ever
Everything above helps you read the situation, but this is the line that doesn't move. Someone you have only met online β no matter how real the photos, how smooth the video, how deep the connection feels β never has a legitimate reason to need your money, your gift cards, your crypto, or a one-time code from your phone. The FTC logged more than $800 million in reported romance-scam losses in 2024 alone, and the request for money is the moment nearly every one of those turned. If it arrives, the relationship was the setup. Stop, and report it.
Red flags worth training yourself to notice
No single item here proves a fraud, but each one earns a real check before you trust the person further β and several together rarely add up to anything good.
- βPhotos that look like a magazine shoot, or that reverse-search back to a model, influencer, or stock library.
- βAlways a reason they can't do a live video call β broken camera, bad signal, "I'm shy on camera."
- βThey push to move off the app fast β to WhatsApp, Telegram, or text β where there's no record.
- βThe story shifts: the city, job, or time zone doesn't line up between conversations.
- βIntense affection far too soon β "love-bombing" that outpaces how little you've actually verified.
- βReverse search comes back completely empty, which can mean an AI-generated face, not a real match.
- βAny request for money, gift cards, crypto, bank details, or a verification/one-time code.
Why reverse image search isn't enough on its own anymore
For years, the standard advice was simple: reverse-image-search the photos and you'll catch the fake. It still works often enough to be the first step β but it now has a blind spot you have to account for. Cheap AI tools can generate a face that belongs to no one, which means there's nothing anywhere for a reverse search to match. A "no results" screen used to feel reassuring; today it's ambiguous. That's the whole reason the later steps lean on live video and liveness tests instead of images β they check for a real human responding in real time, which a generated headshot can't do when you ask it to wave.
The one habit that does most of the work
Slow the relationship down to the speed of verification. Romance scams run on urgency β an emergency, a narrowing window, a reason you must act or decide right now. A real connection survives a pause; it'll still be there next week after you've done a video call and run the photos. A fabricated one depends on you not stopping to look. Whenever the pressure to commit β emotionally or financially β outpaces what you've actually been able to confirm, that gap is the signal.
What this guide won't do
It won't tell you that the specific person you're talking to is real or a fraud β no guide can, without being in the conversation. What it gives you is a repeatable way to find out for yourself, and the clarity to act on what you find. Run the checks where the stakes are real, keep money entirely out of any relationship you've only had on a screen, and when you genuinely can't verify a person, treat that uncertainty as the answer until it's resolved.
Check a photo or a claim in seconds
Not sure about a profile, a story, or a viral claim tied to it? FAXTR searches 100+ fact-checking organizations in one query β free, no login required.