A fraud victim loses $8,000. Easy number to record.
What happens next is messier.
The platform adds another verification vendor. Support hires two people. Fraud analysts review more accounts by hand. Legitimate customers get stopped by new checks and complain. Paid traffic keeps arriving, but a larger share of visitors search the company name with “scam” before they register.
Nobody sends the CFO an invoice marked loss of trust.
The bill still arrives.
Romance fraud gives digital companies a particularly clear view of this problem because the product depends on strangers believing each other. DateMyAge operates in that uncomfortable spot. Its users want connection; scammers need the same channels to look convincing.
The FBI’s 2025 numbers show how expensive that collision has become.
Fraud losses are growing faster than the case count
The FBI’s Internet Crime Complaint Center recorded 23,159 confidence and romance fraud complaints in 2025. A year earlier, it recorded 17,910.
That’s a 29% rise.
Reported losses moved faster, from $672 million in 2024 to $929.3 million in 2025, an increase of 38%.
That difference is the number executives should pay attention to.
More incidents are being reported, but the average damage is also getting harder to ignore. And these are reported cases. The FTC has repeatedly warned that many scams never reach a government agency.
The age split is worse.
People aged 60 and over filed 10,188 confidence and romance fraud complaints with the FBI in 2025. Their reported losses came to $584 million.
In other words, roughly 63 cents of every dollar recorded in that FBI fraud category came from people 60 and older.
For a service built around mature users, this isn’t an abstract cyber-risk sitting somewhere in an IT budget. It reaches the core product.
A customer who has heard three scam stories from friends doesn’t arrive in the same state of mind as a customer who assumes every profile is genuine. They check. They hesitate. They search the DateMyAge reviews about news and safety and a scam before entering payment details.
That search has an economic value, even if Google Analytics doesn’t put a neat dollar amount beside it.
Romance fraud losses rose much faster than complaints
FBI IC3 confidence and romance fraud data for 2024 and 2025.
Complaints
Losses, $M
0600012000180002400020242025
Source: FBI Internet Crime Complaint Center, 2025 Annual Report
The chart is deliberately uncomfortable: complaints increased sharply, but the money climbed even faster.
Trust has become part of customer acquisition
Marketing teams normally talk about CAC in familiar terms.
Media spend. Creative. Landing pages. Conversion rates.
Fraud adds another layer.
Suppose paid social sends 10,000 potential customers to a dating platform. Some arrive ready to register. Others open a second tab and search for complaints, billing problems, fake-profile stories and cancellation issues.
That isn’t a branding problem in the old PR sense.
It’s part of the purchase funnel.
The FTC reported that nearly 30% of people who lost money to scams in 2025 said the contact began on social media. Those reports accounted for $2.1 billion in losses. For romance scams, nearly 60% of people reporting a loss said the first contact happened on a social platform.
A channel that brings business customers also brings criminals targets.
That changes what “frictionless acquisition” means.
You can spend aggressively making registration fast, then discover that the market itself has taught customers to slow down.
SWAGGER recently touched a similar problem in its piece on screening and identity checks in rideshare businesses. The industries are different, but the operating question is familiar: what should a platform know before it allows one stranger access to another?
Dating companies have an extra complication. Intent isn’t printed on an ID card.
A passport can establish that a person exists. It can’t tell you what they’re planning to do after the account goes live.
Six checks a day sounds excessive until you price the alternative
DateMyAge says the first 48 hours of an interaction deserve particular attention because scammers often try to move quickly.
Its response is not a one-time selfie.
According to the company, its machine-learning systems scan conversations for malicious links and fraud signals every four hours, including weekends and public holidays.
Six screening cycles per day.
DateMyAge also says some members go through government-ID checks and biometric liveness verification. Verification status can be reviewed again later rather than treated as permanent.
Among users who attempted verification across DateMyAge and other platforms operated by its parent company, the company reported a 79% decline in fake applicants between 2024 and 2025.
That figure doesn’t mean verification solves fraud.
It does show what modern trust infrastructure actually looks like: repetitive and expensive.
An engineer has to maintain the system. A model flags questionable behaviour. Someone has to decide what happens to the account. False positives land in customer support. Appeals need handling.
The old internet model was cheap: create account, confirm email, enter.
That model looks less attractive once fake identities become cheap too.
The FBI recorded 626 confidence and romance fraud complaints referencing AI in 2025, connected to just over $19 million in reported losses.
That is still a small slice of total complaints. The more interesting issue is cost.
AI makes decent scam content inexpensive.
A criminal can rewrite a message in seconds. Generate a plausible biography. Maintain several conversations without forgetting basic details. Produce images without stealing a stranger’s entire Facebook album.
The platform then spends real money distinguishing those accounts from real customers.
One side’s production cost falls. The other side’s defence cost rises.
Bad trade.
The transaction is almost the end of the scam
Businesses tend to focus on the moment money moves because that’s where the clean number appears.
By then, a romance scam may have been running for weeks.
There was a profile. Conversations. Personal details. Perhaps calls. Then a reason money was suddenly needed.
One 2026 US case showed how far that process can go. The Guardian reported on a federal fraud prosecution involving allegations that a man posed as a wealthy former NFL player while approaching women through dating apps and pitching supposed investments.
That is why transaction monitoring alone is late.
The useful signals often appear before the request for money.
One account sends near-identical messages to dozens of people.
A new user pushing every conversation toward an external messaging app.
Investment language suddenly appearing in romantic chats.
Repeated attempts to share suspicious links.
None of those behaviours proves a crime by itself. Together, they can justify a closer look.
A platform sees something an individual user cannot: patterns across hundreds or thousands of interactions.
That’s the advantage worth paying for.
Good friction can protect revenue
There is a reflex in product management to treat every extra step as conversion damage.
Sometimes that’s correct.
Asking a legitimate user for the same document three times is lousy product design.
Stopping an account whose behaviour suddenly changes is different.
A useful fraud control creates friction around abnormal behaviour rather than spreading irritation across every customer.
That distinction matters commercially.
If verification becomes unbearable, users leave.
If verification is too weak, scammers arrive and users leave for another reason.
The job is not to remove friction. It is to spend it carefully.
And that is where the hidden cost of online scams gets interesting.
The $929.3 million recorded by the FBI is money victims reported losing. It doesn’t include the engineer building another detection rule. The support agent reviewing a suspended profile. The legitimate customer who quits registration because the process suddenly feels intrusive. Or the prospective user who never signs up after reading three fraud stories over breakfast.
No one company created that distrust.
Every company operating in the category gets charged for it.
For digital businesses, fraud prevention has stopped being the department that says no.
It’s becoming part of the cost of making a stranger comfortable enough to say yes.
