Suss
All articles
CatfishingRed flags

How to Spot a Catfish Before the First Date

By Suss Editorial TeamPublished August 26, 2026 15 min read
Illustration of a glowing dating profile on a phone with a hidden mask behind it
In this guide

Catfishing works because it exploits something good in you: your willingness to believe someone. The person on the other side of the chat is funny, attentive, and suspiciously well-matched to you. By the time the excuses start — the cancelled video calls, the sudden emergencies, the crypto opportunity they only mention because they care — you are already emotionally invested, and the part of your brain that would have said no is now negotiating on their behalf.

The good news is that catfish profiles are not subtle once you know what to look at. Nearly all of them share the same fingerprints, because nearly all of them are built the same way: a small set of stolen or generated photos, a story thin enough to keep straight, and a script designed to accelerate intimacy faster than your judgement can keep up. This guide walks through every layer of the check, in the order it makes sense to do it — photos first, because it is the cheapest and highest-yield signal, then story, then behaviour, then the specific scam patterns, then what to actually do with what you find.

Read it once properly and you will have a repeatable routine that takes about ten minutes per match, which is a small price for not spending six months and your savings on someone who does not exist.

Start with the photos: what fake images look like

Photos are where catfish are weakest, because they cannot generate an unlimited, coherent visual history of a person who never lived. Real people leave a messy trail of images: bad lighting, group shots, haircuts that change, a jacket they clearly wore for three winters. Fabricated people do not. Their photo set is small, flattering, and oddly disconnected.

The thin-portfolio problem

  • Only two to five photos total, all flattering, no throwaway shots — nobody's real camera roll looks like that.
  • Every photo is a solo photo. Real dating profiles almost always leak at least one friend, sibling, dog, or wedding.
  • No environmental continuity: no repeated rooms, cars, streets, or clothing across pictures.
  • Photos that look professionally shot or model-portfolio styled, especially if the person's claimed job would never produce them.
  • Wildly inconsistent image quality — one crisp studio photo next to three low-resolution crops, which usually means the crops were harvested from different sources.

Signs an image was AI-generated

Generated faces have improved enormously, but they still fail in predictable places. Zoom in and look at the edges of the image rather than the face itself — the face is the part the model is best at.

  • Hands: extra or fused fingers, jewellery that merges with skin, a hand that does not quite hold the object it is holding.
  • Teeth and ears: too many teeth, uneven ear shapes between photos, earrings on one side only.
  • Backgrounds that warp: text on signage that dissolves into nonsense, straight lines that bend, patterns that repeat unnaturally.
  • Glasses, collars, and hairlines that blend into the background instead of ending cleanly.
  • Skin that is uniformly smooth with no pores, moles, or asymmetry — and a face whose bone structure shifts subtly between images of 'the same' person.
  • Perfectly centred composition with shallow depth of field in every single shot, which is a strong tell of a generator default rather than a phone camera.

How to run a reverse image search properly

This is the single highest-value five minutes you can spend. Most people do it wrong by uploading one photo to one service and stopping when nothing comes back.

  • Screenshot and crop each photo to the face only, then search each crop separately — backgrounds and body poses confuse matching.
  • Run every photo, not just the main one. Scammers often mix one stolen photo of a real person with several generated ones.
  • Use more than one engine. Google Lens, Bing Visual Search, Yandex, and TinEye return meaningfully different results, and Yandex in particular is strong on faces.
  • Search for the person's claimed name in quotes alongside their city and job. A real adult with the job they describe usually leaves at least one trace.
  • Note the negative result correctly: finding nothing does not prove the photos are real, it only means they are not indexed. Absence of evidence is not evidence of honesty.

The live-photo test

The cleanest test costs nothing: ask for a casual, specific photo taken now. Not a selfie — anyone can take a selfie from a folder of stolen selfies. Ask for something a stolen archive will never contain, like holding up three fingers, or standing next to a specific object, or their coffee cup with today's newspaper. A real person finds this mildly silly and does it in thirty seconds. A catfish stalls, negotiates, gets offended, or sends something that does not match the request.

They will not video call — ever

One cancelled call is life. A pattern of cancelled calls is a signal, and it is the most reliable one in this entire guide. Deepfake video is real but still expensive, awkward, and rare in volume romance scams; the overwhelming majority of catfish simply avoid the camera and manage your expectations around it.

The excuses are so consistent they are almost a script: a broken front camera, terrible connection on an oil rig or military base, a job that travels constantly, deep shyness about their appearance, a religious or cultural reason, or a promise to call 'next week' that renews weekly. Any single one of these can be true. All of them arriving in sequence, from someone who is otherwise extremely available to text you for hours, is not a coincidence.

How to run the call test without an argument

  • Ask early — within the first few days — while there is no emotional cost to a no. The longer you wait, the more you will rationalise.
  • Keep it low-stakes: two minutes, camera on, no makeup required, any app they like.
  • Watch for live behaviour, not just a face: do they respond to something you say in real time, look where you point, hold up a hand when you ask?
  • Be alert to a call that is oddly static — frozen framing, a loop, audio that never quite syncs, or a person who will not turn their head.
  • Do not accept a pre-recorded video as a substitute. It proves a video exists, not that they are in it.

If weeks have passed and you have never seen them move, speak, or react live, treat that as the answer. You do not need a confession.

The story keeps moving

Fabricated lives fail under repetition. A real person's biography is boring and stable: the same employer, the same sister, the same commute, the same complaints. An invented one drifts, because the person maintaining it is running several conversations at once and remembering none of them perfectly.

What inconsistency actually looks like

  • Job details that shift: an engineer who later describes work no engineer does, or a company name that changes spelling.
  • Geography that does not add up — timezone-inconsistent messages, weather that does not match their claimed city, local landmarks they cannot name.
  • Family details that mutate: one child becomes two, a late spouse's story changes cause or date.
  • Language patterns that swing between messages, suggesting more than one person is typing, or heavy reliance on translated-sounding phrasing that appears only sometimes.
  • Vagueness that is strategic rather than shy: they will talk for an hour about feelings and never once name their street, employer, or a friend.

The cheap questions that expose it

You do not need to interrogate anyone. You need a handful of ordinary, unremarkable questions that a real local answers instantly and a script cannot:

  • Which supermarket or coffee place is nearest you? What's it like?
  • What's the worst part of your commute?
  • What did you actually do at work today — like, the boring version?
  • Who did you last have dinner with?
  • Which neighbourhood should I avoid living in around there?

Note the texture of the answer, not just the content. Real answers are specific, mildly annoyed, and full of irrelevant detail. Fabricated ones are generic, warm, and immediately redirect to you.

They rush intimacy: the love-bombing engine

Love bombing is the fuel of almost every romance scam, and it is what makes otherwise careful people ignore everything above. It looks like intense, sustained attention: good morning and good night messages from day two, talk of fate and soulmates within a week, pet names before a phone call, future plans before a meeting, and a level of emotional availability no employed adult sustains naturally.

It feels like being chosen. That is precisely the point. Manufactured intensity builds a bond fast enough to bypass evaluation, and it also builds obligation: once someone has told you that you are the best thing that has happened to them, refusing them anything feels cruel. That obligation is the product being manufactured.

Healthy interest vs manufactured intensity

  • Healthy interest is curious about your actual life; love bombing is complimentary about you as an idea.
  • Healthy interest respects pace; love bombing frames your caution as coldness.
  • Healthy interest survives a boundary; love bombing punishes one, with sulking, silence, or escalation.
  • Healthy interest gets more specific over time; love bombing stays at the same emotional volume regardless of how little you actually know each other.

A useful private rule: emotional intimacy should never outrun verified reality. If someone is talking about forever and you have never seen their face live, that gap is the risk.

They move you off the app fast

Dating apps have moderation teams, reporting flows, photo checks, and automated scam detection. Every one of those is a threat to a catfish, which is why the push to move to WhatsApp, Telegram, Google Chat, or email usually arrives within the first few messages — often before you have exchanged anything of substance.

Moving platforms eventually is normal. Being rushed there immediately is a tell, and there is a second reason it matters: off-app, there is no record for the platform to act on and nothing to report. You lose the safety net exactly when you are most likely to need it.

  • Keep the conversation in-app until you have had a live video call.
  • If you do move, keep screenshots of the original profile and the early messages.
  • Be wary of requests to switch to a platform with disappearing messages by default.
  • Never move to a platform they insist on for reasons that are about secrecy rather than convenience.

The money moment: how romance scams actually close

Catfishing is rarely the end goal. It is the setup for an ask, and the ask follows a small number of well-worn scripts. Knowing them in advance is protective, because the whole design relies on the request feeling like a one-off emergency rather than a step in a process.

The common scripts

  • The emergency: a medical bill, an accident, a detained relative, a customs fee on a gift they sent you. Always urgent, always solvable with a transfer.
  • The blocked-funds story: they have money but cannot access it — a frozen account, a contract in escrow, a wallet they need help withdrawing from.
  • The investment mentor ('pig butchering'): they are not asking for money, they are generously showing you a crypto or trading platform where they have done well. Small withdrawals work at first. Larger ones do not.
  • The travel-to-meet ask: they desperately want to visit you and are just short on the ticket, then the visa, then the deposit.
  • The intimate-images setup: warmth escalates to explicit exchange, then to blackmail. This is sextortion, and paying never ends it.

Payment methods that should stop the conversation

Regardless of the story, the requested method tells you almost everything. Scammers need irreversibility.

  • Crypto transfers of any kind, including 'just to a wallet I set up for you'.
  • Gift cards, prepaid cards, or reading card codes out over chat.
  • Bank wires to a third party, or to a name that does not match theirs.
  • Peer-to-peer apps used as 'friends and family' with no buyer protection.
  • Anything at all where you are asked to keep it secret from your family or bank.

Your ten-minute pre-date verification routine

Put the whole thing together and it is short enough to run on every match, which is the only way any of it actually helps. Habits protect you; one-off vigilance does not.

  • Crop and reverse-search each profile photo across two engines. (3 minutes)
  • Search their name in quotes with their city and claimed employer. (2 minutes)
  • Ask two cheap local questions and read the texture of the answers. (1 minute)
  • Request a two-minute video call before you agree to meet. (1 minute to ask)
  • Scan the conversation for pace: has intimacy outrun verification? Has money, crypto, or a platform switch come up at all? (2 minutes)
  • Tell one friend who you are talking to, and send them the profile. (1 minute)

If you would rather not run all of this manually, that is exactly what Suss is for. A check reads photo signals, public profile signals, and behaviour patterns from your conversation in about 90 seconds, and presents them with their limits stated plainly — so you get a calmer read without turning yourself into an investigator every time you match with someone.

What to do with a bad feeling

You do not need proof to slow down, and you do not owe anyone a confrontation. Confronting a scammer achieves nothing useful: they will not confess, they will escalate emotionally, and they will use the accusation to reframe you as the one who did harm. There is no version of that conversation that leaves you better informed.

If you are still talking to them

  • Stop sharing new personal details — address, workplace, financial situation, family names.
  • Do not send money, and do not accept money or packages to forward. Both make you a participant in fraud.
  • Do not send intimate images. If you already have, stop, and do not pay any demand.
  • Screenshot the profile and conversation before you block, so there is something to report.
  • Block, then report the profile in the app and to your national fraud reporting body. Reporting is what protects the next person.

If you have already lost money

  • Contact your bank or card provider immediately and use the words 'authorised push payment fraud' or 'scam' — speed materially affects recovery odds.
  • Report it to your country's fraud authority and to the platform where you met.
  • Expect a recovery scam: someone will contact you offering to retrieve your funds for a fee. That is the same crew or their friends.
  • Tell someone you trust. Shame is the mechanism that keeps victims paying, and it is the first thing to dismantle.

None of this is about becoming cynical. The overwhelming majority of people on dating apps are exactly who they say they are, badly lit and slightly nervous. A short verification habit does not make you paranoid — it makes the real ones easier to enjoy, because you are no longer quietly wondering.

Frequently asked questions

What is the fastest way to check if someone is a catfish?
Run their clearest profile photo through a reverse image search and ask them for a video call. A real person with nothing to hide can usually do one of those within a day; a catfish will stall, deflect, or produce excuses for both. Together those two tests catch the large majority of fake profiles in under ten minutes.
Can a catfish pass a video call?
It is getting easier with real-time deepfake tools, but it is still rare and usually detectable. Watch for frozen or looping video, audio that lags the lips, a caller who keeps the camera at odd angles, or constant 'connection problems' whenever you ask them to do something specific like wave or hold up fingers. Suggest a spontaneous call rather than a scheduled one — preparation is what makes fakes convincing.
How can I tell if a profile photo was made with AI?
Zoom in on the details: ears, teeth, hands, jewellery, and the background. AI images often have asymmetric earrings, blurred or melted background text, overly smooth skin with no pores, and lighting that does not match across the face. Also check whether every photo looks like it was taken in the same session — real people have inconsistent camera rolls.
Is one red flag enough to conclude someone is fake?
No. Single signals have innocent explanations — some people genuinely hate video calls or have a thin social media footprint. What matters is the stack: refusing calls, plus moving you off the app fast, plus photos that reverse-search to someone else, plus rushing intimacy. Two or three strong signals pointing the same way is when you act.
What should I do if I realise I have been talking to a catfish?
Stop replying, screenshot the profile and conversations, block them, and report the account to the app. If money or explicit images are involved, also report to the FTC (or your national equivalent) and your bank immediately. Do not confront them expecting an admission — the script for being caught is just another script.

Frequently asked questions

Ask for a short live video call in the first few days. Deepfake video is still rare in volume romance scams, so most fake profiles simply avoid the camera and manage your expectations with excuses about broken cameras, bad connections or travel.

No. Finding a match proves the photo was taken from somewhere else, but finding nothing only means the image is not indexed. Crop each photo to the face, search every image separately, and use several engines — Google Lens, Bing, Yandex and TinEye return different results.

Look away from the face. Generators still fail on hands, teeth, ears, jewellery, glasses frames and text in the background, and they produce uniformly smooth skin with identical shallow-depth composition in every shot.

Treat the reaction as the answer. Anger, guilt-tripping or negotiation over a harmless request — a two-minute call, a photo holding up three fingers — tells you more than the photo would have. Stop sharing personal details and do not send money.

No. A basic check on someone you are already talking to is normal safety behaviour, and honest people are relieved by it. You are confirming that the person you meet is the person you have been talking to.

Check them before you meet them.

Suss reads photo, behaviour, and profile signals on the person you're about to meet — in about 90 seconds.

Upload A Photo

Related guides

More from Catfishing & fake profiles.