The Online Dating Scam Red Flags That Matter Most Right Now

The red flag that counts now is not what he looks like. It is what he does when you ask to see him live, and what he does about money.

That sounds obvious until you notice how much advice still tells you to squint at a profile photo. The FTC's 2025 figures put reported romance scam losses at about $1.48 billion, up 22% on the year before, with cryptocurrency now the leading payment method at roughly 34% of losses, ahead of gift cards at 17%, wire transfers at 15% and bank transfers at 14%. Those are self-reported numbers from people who filed a complaint, so the real total is higher, probably by a lot. Most people who lose money to a man they thought loved them never tell anyone, let alone a federal agency.

It runs in a fixed order, and the order is the point.

First contact usually is not where you think. Nearly 60% of people who reported losing money to a romance scam in 2025 said it started on social media, not a dating app, according to the FTC's April 2026 data spotlight on the $2.1 billion lost to social-media-originated scams. A friend request, a comment, a message from someone who seems to know someone you know.

Then the conversation moves to a private messaging app, off the platform that could flag it. Then weeks of attention, tailored to what your profile told him about you. Then trust is manufactured, sometimes by giving money rather than asking for it. Then the crisis, or the investment opportunity, and the money goes out in crypto.

What artificial intelligence changed is the cost of the first three steps. In a network of roughly 28 dating apps investigated by The Verge and Anthropic, more than 4,700 fabricated personas sent 2.36 million messages to over 25,000 real people in a two-week sample in April 2026. That works out to about 94 messages per targeted person over fourteen days, roughly seven a day. Attentive. Consistent. Never tired. A human con artist could not carry forty conversations at that pitch; software can carry five hundred.

In that same app network, only about one in four apparent matches was an actual person, and the people were generally paid gig workers rather than genuine daters. They were inserted deliberately, so the operation could survive video calls and social media checks. Read that twice. The video call was not a flaw in the scam. It was a feature of it.

The nearly $1 million LinkedIn case reported this year shows the older version and why it worked. The profile used stolen photos of a real man and appeared connected to a former colleague of the victim. The first request was $10,000 to unfreeze a bank account, and the scammer sent it back. "I started to trust him because he sent me back money," she said. Requests then climbed to $20,000, $40,000, around $200,000 for customs fees, and finally 500,000 for surgery, framed as proof of love. Meetings were cancelled, video calls dodged, for two years.

And it is organised. Six Nigerian nationals linked to the Black Axe network were extradited to the United States on 11 September 2026, accused of taking more than $6 million from over 100 American women. A month earlier, a Georgia man was convicted of laundering more than $2.7 million of romance scam proceeds through a car dealership, with at least 35 victims nationwide. The man messaging you may not be the man taking your money, and neither of them is improvising.

Almost nothing, now. If a face was generated rather than stolen, it has never appeared online, so the search comes back empty and the empty result reads as reassurance. Social Catfish, an identity verification firm and therefore an interested party, estimates that about 25% of fake dating profiles in 2026 use fully AI-generated photos rather than someone else's. Treat the figure as a direction of travel rather than a precise count, but the mechanism is not in dispute.

There is a fair argument on the other side, and it deserves saying: most awkward men are just awkward. A man who is clumsy in his first three messages, who has one decent photo and one taken by his teenager in a car park, who says the wrong thing about his divorce, is behaving exactly like a nervous 44-year-old. Awkwardness is not a red flag. Smoothness plus urgency is.

This is the part of the decision you control, and it is worth making on purpose rather than by which app your sister uses.

Match Group has been pushing facial liveness verification hardest. Tinder's Face Check asks new users for a short video selfie, confirms a live person is present, matches the face to their photos and checks whether that face is running other accounts. In markets where it was already mandatory, the company reported more than a 60% drop in user exposure to potential bad actors and more than 40% fewer bad-actor reports. Face Check was mandatory for new users in Canada, Australia, India, Colombia, parts of Southeast Asia and California as of October 2025, with more US states following and a rollout to the rest of the group's apps, including Hinge, Match and OkCupid, planned through 2026. Until that lands, Hinge has no ID verification at any tier, meaning nothing in the app confirms a person's real-world identity.

Bumble runs selfie-based photo verification and identity verification flows introduced in 2025, plus its Deception Detector system, which in 2024 testing auto-blocked 95% of accounts flagged as spam or scam and cut user reports of fake profiles by 45% within two months. Worth knowing that Bumble's February 2026 product announcement was about AI photo and bio feedback, not deepfake detection, so the anti-fraud work there is older than the current threat. On eHarmony I have no verification data in front of me, which is itself a reason to check its safety page before you pay.

None of this makes an app safe. It makes faking a profile more expensive, which pushes the volume operations somewhere cheaper. That is the whole game: be where the con costs more to run.