The statistics on deepfake fraud and romance scams are alarming on their own. What makes them worse: they're almost certainly undercounted. Here's what the data actually shows — and what it means in practice.
Most people who get defrauded by a romance scam or deepfake operation don't report it. The embarrassment of being deceived — of having trusted someone who turned out to be entirely fictional — is too much. Victims often feel that reporting won't help anyway. They're usually right that the money is gone. What they don't realize is that every unreported case lets the operation keep running.
This means that every number in this article is a floor, not a ceiling. The actual scale of deepfake fraud is larger. The numbers are still worth knowing, because even the reported figures are striking.
The Federal Trade Commission tracks consumer fraud reports by category. Romance scams — which include catfishing, pig butchering, and any fraud built on a fabricated romantic relationship — have become one of the highest-loss categories in their database.
$700 million in reported romance scam losses in 2024. That's a figure from people who filed reports with the FTC. Academic researchers and advocacy groups estimate the real number is 3-5x higher, based on victim surveys about whether they reported. If that multiplier is even roughly accurate, actual losses in 2024 were over $2 billion in the US alone.
The median individual loss in 2024 was $2,000. But the distribution is extremely skewed. The median is pulled down by a large number of small losses. The average loss — pulled up by the large pig butchering cases — is much higher. Pig butchering victims routinely lose $50,000-$500,000, with some cases exceeding $1 million.
Adults over 60 report the highest total losses, though younger adults have higher rates of being targeted. The demographic with the fastest-growing loss figures: people in their 30s and 40s who are financially established enough to have money to invest and emotionally available enough to be receptive to a relationship.
Sumsub, an identity verification company, published identity fraud data in their 2024 annual report that included deepfake-specific metrics.
257% increase in deepfake incidents in 2024 compared to 2023. That's not a typo. It's also not a number produced by a small sample — Sumsub operates at the scale of major financial and verification platforms. The growth rate reflects both the improving quality of generation tools and their dramatically lower cost and friction.
The nature of the incidents is worth understanding. The 257% figure isn't primarily about sophisticated video deepfakes used to impersonate executives in board meetings. A significant portion is profile photos — AI-generated headshots used to create fake accounts on dating apps, social networks, and professional platforms. The technology barrier for this use case is essentially zero: free tools generate a convincing face in seconds.
Onfido, another verification firm, reported in their 2024 fraud report that AI-generated ID document fraud was up 31x in three years. The two data points together — face generation and document generation — describe a fraud ecosystem that's gotten dramatically cheaper and more accessible.
Meta (Facebook and Instagram) publishes fake account removal data in its quarterly transparency reports. In Q2 2023, Meta removed 827 million fake accounts in a single quarter. That's one quarter. That figure has stayed in a similar range across subsequent quarters. It represents accounts that were caught — not the ones that weren't.
LinkedIn removed 95 million fake accounts in 2023 across two waves of enforcement. Their 2023 Transparency Report noted that 99.1% of spam and scam content was detected before users reported it — which sounds reassuring until you consider the absolute scale. 0.9% of the content that got through on a platform with a billion users is still a lot of fake content.
Dating apps don't publish comparable transparency data, which is part of the problem. Tinder's parent company Match Group has faced regulatory pressure in multiple countries to provide clearer data on fake account rates. Independent researchers have published estimates ranging from 10% to 25% of profiles on major apps depending on methodology and platform. None of those numbers have been disputed with hard counter-data.
The AI-generated photo shift matters enormously for these figures. Platforms detect fake accounts primarily through behavioral signals — automation patterns, message templates, suspicious activity. AI-generated photos attached to accounts run by real human operators (including, in large operations, trafficked forced labor) don't trigger behavioral detection. The photo check is the gap that platform moderation wasn't built to close.
Pig butchering deserves its own section because the numbers are specifically alarming and it's still not well understood by the general public.
The FBI's Internet Crime Complaint Center (IC3) reported that cryptocurrency investment fraud — which includes pig butchering as its dominant variant — generated $3.96 billion in US losses in 2023, making it the highest-loss category of all internet crime they track. Romance-initiated investment fraud makes up a substantial portion of that total.
The United Nations Office on Drugs and Crime estimated in 2023 that pig butchering operations in Southeast Asia (primarily Myanmar, Cambodia, and Laos) were generating $7.5 billion annually in global losses. Other estimates run higher. The Global Anti-Scam Organization, which works with victims, puts the figure above $10 billion annually when accounting for unreported cases.
The operations behind these numbers are large and professionalized. They're not individual bad actors — they're organizations with hundreds or thousands of employees, management structures, scripts, and training programs. Some of those employees are victims themselves, people who were trafficked into the operations under false pretenses and coerced into running accounts. The U.S. Department of State has formally classified some of these operations as human trafficking situations.
The idea that romance scam victims are uniquely naive or vulnerable is both wrong and harmful. It's wrong because the data doesn't support it. It's harmful because it stops victims from coming forward and reporting.
FTC data on romance scam victims shows:
Higher income = higher losses. People with higher incomes lose more money when they're targeted — not because they're more vulnerable psychologically, but because they have more to lose and the scammer has reason to invest more time in them.
Education doesn't protect. Survey data from academic researchers finds that educational attainment has no significant correlation with vulnerability to romance scams. People with graduate degrees fall for them at similar rates to people with high school diplomas. The scripts are designed by professionals and tested across thousands of targets. They're genuinely hard to resist once the emotional architecture is in place.
Life transitions increase risk. Divorce, bereavement, relocation, retirement — any transition that disrupts existing social networks and creates genuine emotional need makes someone more receptive to a relationship that fills that gap quickly. This is a human pattern, not a character flaw.
The profile that emerges from the data is: someone going through a transition, looking for connection, who encounters what feels like a genuine relationship — because it's designed by professionals to feel like one. That's a profile that fits a very large number of people at various points in their lives.
Every statistic in this article is an undercount. That's not a caveat — it's a structural feature of how fraud data is collected. The FTC, the FBI, and every other agency that tracks romance scam losses depends on victims filing reports. Most don't.
The reasons are consistent across surveys: embarrassment, the belief that reporting won't result in recovery, uncertainty about where to report, and in some cases, not recognizing until much later that what happened to them was fraud at all. One survey of romance scam victims found that fewer than 30% had reported to any authority. If that figure is even roughly accurate, it means every aggregate number in this article represents less than a third of the actual problem.
This matters for how you read the data. When the FBI's IC3 says cryptocurrency investment fraud caused $3.96 billion in US losses in 2023, that's not "the number." That's the floor. The FTC's $700 million in romance scam losses is a floor. Sumsub's 257% growth figure is based on incidents that were submitted through their verification systems — companies that integrated their API and reported the results. The actual volume across all platforms is higher.
The point isn't to be alarmist. It's to calibrate correctly. If the reported numbers are this large, the actual scale of the problem is large enough that essentially everyone using dating apps, social media, or online marketplaces should treat photo verification as a standard habit, not an exceptional precaution.
The 257% growth rate in deepfake incidents isn't an anomaly — it reflects an underlying trend that hasn't plateaued. The cost of generating convincing AI faces has dropped from specialized tools requiring technical skill to free web apps that produce results in seconds. The quality has improved simultaneously. The operations using these tools have gotten more professionalized.
Platform detection has improved too, but it's chasing a moving target. The gap between generation capability and detection accuracy has generally stayed consistent or widened. New generator architectures produce images with cleaner signatures that existing detectors miss until they're retrained. The lag can be months.
User-level detection tools — browser extensions that analyze images at the point where you see them — matter because they put the check in the hands of the person who needs it, rather than depending on platforms to catch everything before it reaches you. Platforms haven't solved this and show no signs of solving it in the near term. The data makes that pretty clear.
For a practical guide to protecting yourself, see the romance scam prevention guide. For the mechanics of how scam operations run, see the article on how catfish scams work. For checking profiles now, the catfish detector page covers how Faux Spy works.
10 free checks per day. Works on any website in Chrome.
🕵️ Add to Chrome — Free