Romance Fraud Targets Gen Z: Why Platforms Should Do the Checking

For a decade the romance scam has had a face, and it was grey. The widow who wired her pension to an oil-rig engineer she never met. The retired teacher who remortgaged for a soldier stranded somewhere hot and dangerous. Banks designed their warnings for that person. Charities built their leaflets around her. The rest of us, especially the young, filed the whole genre under things that happen to people who did not grow up online.

On 25 September 2026, figures from Lloyds Banking Group landed that should retire that picture for good. Reports of romance scams to the bank rose 24 per cent in the twelve months to the end of August 2026, compared with the year before. Within that rise, reports from victims aged 18 to 24 jumped 59 per cent. Those from victims aged 25 to 34 climbed 39 per cent. The older victims have not gone anywhere: more than half of the money lost, 55 per cent, still came from people over 65, and victims aged 75 to 84 lost an average of £9,051 each, the highest of any age group and 19 per cent more than a year earlier. But the fastest growth is now among people who were born with a phone in reach. According to Lloyds, those younger victims typically met the person defrauding them on dating apps and social media, while the over-50s were more often approached through traditional dating websites and social platforms.

In the same week, two economists at Central South University in Changsha posted a short paper to arXiv with a title that reads like a diagnosis. “When Trust Attracts Fraud: AI and Trust Arbitrage”, by Xieyu Yin and Fenghua Wen and linked to the journal Economics Letters, opens with a sentence worth pinning above every trust and safety team's desk: “Trust can attract fraud when it delays verification.” Their model says nothing about dating. It is a piece of abstract economics about sellers, signals and markets. But read it alongside the Lloyds figures and it describes, with uncomfortable precision, what is happening to a generation that was supposed to be too savvy to be caught.

The two releases together raise three questions. Why is the fastest-growing group of victims the one everyone assumed was immune? What does a flood of synthetic faces, voices and photographs do to the ordinary, decent habit of giving a stranger the benefit of the doubt? And if trust attracts fraud by delaying the moment anyone checks, whose job is the checking: the person swiping at midnight, or the companies whose platforms the fake suitors are built on?

The Generation That Was Supposed to Know Better

The assumption that young people are scam-proof rests on a confusion between fluency and scepticism. Someone who can spot a phishing text from a fake courier is fluent. That is not the same as being sceptical of a person who has spent three weeks being kind to you. And the evidence that the young were never as protected as their reputation suggested predates the Lloyds data. In September 2023, UK Finance and its Take Five campaign published a OnePoll survey of 2,000 UK adults. Young adults aged 18 to 24 were the most confident of any age group in their ability to spot a scam: 91 per cent said they were confident they could identify a fake request for personal information online. Yet just 27 per cent said they would always take steps to check whether an unexpected contact could be trusted, against more than 60 per cent of people over 55. Almost half of the 18 to 24 year olds, 49 per cent, said they had been contacted by an impersonation scammer, and of those targeted, 52 per cent said they had shared personal information or made a payment as a result.

That survey was about impersonation, not romance, and it is three years old. But its pattern, high confidence and low checking, is the one the romance data now reflects. American figures agree. The Federal Trade Commission has long reported that younger people who file fraud complaints are more likely than older people to say they lost money, while older victims report far larger individual losses. In one FTC analysis, 51 per cent of complaints from consumers aged 19 and under reported a financial loss, compared with 21 per cent of complaints from people aged 80 and over, although the median loss for the youngest group was far smaller.

The Lloyds numbers sit neatly on that fault line. Young people are showing up more often; older people are losing more per case. The average loss across all romance scam victims at Lloyds actually fell by 20 per cent to £4,078, which is what you would expect if the victim pool is being swelled by people with smaller bank balances. The fraud is not getting gentler. It is getting wider.

What the Numbers Do and Do Not Say

Some honesty about those percentages is in order. The Lloyds data counts reports to one banking group, not the whole country. A rise in reports is not automatically a rise in crime; it can also reflect victims feeling less ashamed to come forward. A 59 per cent rise from a small base can represent fewer people than a modest rise in a large one, and Lloyds did not publish absolute case numbers by age in the material reported by the press. So the figures are directional. But the direction is corroborated elsewhere. UK Finance's Annual Fraud Report 2026, published in June and covering the calendar year 2025, found that losses to romance fraud across its members rose 23 per cent to £39.2 million, within a wider picture of £1.28 billion stolen through payment fraud. The report described romance scams as relying less on technical intrusion and more on persuasion, emotional manipulation and the exploitation of trust, delivered at scale through digital channels.

In the United States, the FTC's April 2026 data release on social media scams found that nearly 60 per cent of people who reported losing money to a romance scam in 2025 said it started on a social media platform. Scammers, the agency noted, often tailored their pitch to people's profiles, later “inventing a crisis requiring money or casually offering investment advice”. That last phrase describes the hybrid now dominating the field: the romance that turns, gradually, into a cryptocurrency “opportunity”.

Confidence Is Not a Defence

Why would digital natives be caught? Part of the answer is exposure: fraud follows attention, and attention is on the apps. But the deeper answer lies in how trust itself works, and here the psychology is long established. Timothy Levine, a communication scholar whose work on deception has shaped the field, set out truth-default theory in the Journal of Language and Social Psychology in 2014. Its central claim is that people tend to believe others by default, and that this default is adaptive. Most communication is honest. We abandon the truth default only when something triggers suspicion strongly enough, and even then we are poor at spotting lies from demeanour alone.

Romance fraud is engineered to keep that trigger from firing. Monica Whitty, a psychologist who has studied the online dating romance scam for more than a decade, developed what she called the Scammers Persuasive Techniques Model in a 2013 paper in the British Journal of Criminology, based on interviews with victims. The model traces a sequence of stages: a carefully built profile, a grooming period in which intimacy deepens, a first small request, then escalation. Whitty also pointed to the “near-win” phenomenon, borrowed from gambling research, to explain why victims stay in the scam and why some are defrauded again. Each request comes with the promise that this is the last hurdle before the relationship becomes real.

None of those mechanisms care about digital literacy. Knowing how a phishing link works does nothing to protect you from someone who has spent weeks asking about your day, remembering your sister's name and sending good-morning messages at exactly the right hour. Confidence may even make things worse. If you believe scams look like clumsy emails from a prince, a warm, fluent, attentive stranger does not match the template, so the alarm never sounds. Liz Ziegler, fraud prevention director at Lloyds, put it plainly in the bank's release: “Romance scammers don't care how old someone is. They look for opportunities to build trust, create an emotional connection, then turn that relationship into a request for money.”

There is a cultural dimension too. For many people under 30, meeting a partner online is simply how dating happens. That normalisation is healthy, but it has removed the old social friction, the sceptical friend or the raised eyebrow, that once surrounded romance with someone you had never met. A generation that treats online connection as real connection has, reasonably, extended its truth default to the people it meets there.

A Person Now Costs Almost Nothing to Make

Now add generative AI. For most of the history of online romance fraud, the hardest part of the con was being convincing for long enough. Scammers stole photographs from real people's social media accounts, often soldiers, doctors or attractive models, and the countermeasure was simple: run a reverse image search and see whether the face belongs to someone else. Lloyds still recommends exactly that, alongside watching for intense declarations of love, newly created profiles, pressure and guilt, and talking to someone you trust before sending money.

The reverse image search, though, assumes the photograph exists somewhere else. A synthetic face does not. It can be generated fresh, consistent across dozens of images, placed in plausible holiday settings and aged or restyled on demand. The same collapse in cost applies to voice and video. In September 2024, Starling Bank published research, conducted with Mortar Research across more than 3,000 UK adults, warning that fraudsters could clone a voice from as little as three seconds of audio. More than a quarter of respondents said they had been targeted by a voice-cloning scam in the previous year.

Live video, long treated as the definitive proof that someone is real, has gone the same way. In October 2024, Hong Kong police arrested 27 people linked to a syndicate operating from an industrial unit in Hung Hom that allegedly used AI face-swapping to pose as attractive women on video calls, drawing victims across Asia into fake cryptocurrency investments. Police put the losses at around HK$360 million, roughly US$46 million. Earlier that year, WIRED reported, drawing on videos gathered by David Maimon, a criminologist at Georgia State University and head of fraud insights at SentiLink, that West African scammers known as Yahoo Boys were running real-time face-swaps on video calls with little more than two phones and off-the-shelf apps.

At industrial scale, the automation is even starker. In July 2026, the United Nations Office on Drugs and Crime published a threat assessment on South-East Asia's criminal economy. It estimated scam losses across East and South-East Asia, Australia and New Zealand at between $88.3 billion and $114.1 billion in 2025, and documented the use of generative AI and deepfakes in automated fraud. People from at least 80 countries and territories have been identified inside scam compounds in the region, many recruited under false pretences and held in conditions UNODC describes as human trafficking for forced criminality. Delphine Schantz, UNODC's regional representative for South-East Asia and the Pacific, described the model bluntly: “Their operating model looks like corporate franchising.”

That is the industrial base behind the investment-romance hybrid often called pig butchering. In August 2025, Meta said it had removed 6.8 million WhatsApp accounts linked to criminal scam centres in the first half of the year, and disclosed working with OpenAI to disrupt a Cambodian network that had used ChatGPT to generate outreach messages. A human scammer in a compound could once manage a handful of conversations. With a language model drafting replies, translating fluently and remembering every detail of a target's life, the limit on how many “relationships” one operator can sustain begins to dissolve.

Here the Yin and Wen paper earns its place. Their model has three costs that generative AI pushes down at once: fabrication, making the fake face, voice and biography; targeting, reaching the right people; and verification, checking. They have not fallen at the same speed or for the same people. Fabrication and targeting have collapsed for fraudsters with industrial tools. Verification has become cheaper only for whoever actually holds the tools to do it.

The Benefit of the Doubt Was Always a Subsidy

What happens to the habit of giving strangers the benefit of the doubt in this environment? The honest answer is that it becomes more expensive, and the cost falls on the people who keep extending it.

Think of the benefit of the doubt as a subsidy. When you assume a new match is who they say they are, you save yourself the effort of checking and spare them the indignity of being checked. Across millions of interactions that saving is part of what makes online dating work at all.

The subsidy holds up as long as fakes are rare and costly to make. Truth-default theory depends on base rates: the default is adaptive because most people, most of the time, are telling the truth. When a stranger's face, voice and video can all be manufactured, the base rate on some platforms shifts, and the old cues that used to justify trust, a consistent set of photographs, a warm voice note, a live video call, stop carrying information. They were never proof. They were evidence that faking was costly. That is no longer true.

There are two bad ways this can go. People can keep extending the old trust to cues that no longer mean anything, and a growing minority are defrauded. Or they can withdraw trust wholesale, which would be a real loss, because trust is not a flaw to be engineered away; it is the basis of cooperation. A generation taught to regard every warm stranger as a probable criminal would be protected from romance fraud in roughly the way someone who never leaves the house is protected from traffic.

The more useful question is not whether to trust but where the checking should happen, and at what point in the relationship. That is exactly the question the arXiv paper addresses.

Trust Arbitrage and the Market That Forgot to Check

Yin and Wen build a model with two markets that differ in their “prior quality”, roughly how honest the average participant is believed to be. Honest sellers provide genuine evidence cheaply; dishonest ones fabricate it at a cost; buyers can pay to verify. As AI capability grows, fabrication, verification and targeting all get cheaper.

Their first result is what they call a trust valley. If fabrication becomes profitable before verification becomes worthwhile, credibility falls, because evidence that used to mean something can now be faked. It recovers only once checking becomes cheap enough to start. Their second result is the uncomfortable one. A market with higher prior quality delays verification, because when most participants are honest, checking feels like a waste. The better the reputation, the longer everyone waits before bothering to look.

Put two markets side by side and a window opens. In that interval, the lower-quality market has started checking, but the higher-quality one has not. If targeting is cheap enough, deceivers move into the reputable market precisely because it feels no need to check. In the words of the abstract, “deceptive sellers enter the higher-quality but less vigilant market, and their entry can initially reverse its reliability advantage.” The authors call this trust arbitrage. They also stress that it is self-limiting: the inflow of fraud eventually triggers verification in the trusting market too, and the gap closes. They describe the result as “an endogenous but temporary protection gap that redirects deception across markets”.

The paper is a formal model, not an empirical study of dating, and it would be overreaching to treat it as proof of anything about Tinder or Hinge. But as a lens it is sharp. For years, the market that did its checking was the one full of warnings: older people, their banks, charities, and the families who had been told to watch for Nigerian princes and stranded soldiers. The market that felt no need to check was the one everybody assumed was sophisticated: the young. By the logic of the model, that is exactly the market fraud should flow towards once fabrication and targeting become cheap. And the Lloyds data looks very much like an inflow.

The model's comforting conclusion, that the gap is temporary, deserves scrutiny. In the paper, the market that was fooled starts checking. In reality, the “correction” happens one broken bank account and one broken heart at a time. A temporary protection gap in an economics paper is a permanent loss for the people who fall into it. The policy question is whether we wait for that gap to close through victimisation or close it deliberately, earlier, by moving verification to where it is cheapest.

Why Young Users Sit in the Protection Gap

Mapped onto the Lloyds data, the model suggests young people are exposed not because they are foolish but because they sit where checking has been culturally postponed.

Older people have been the target of sustained awareness campaigns for years, and adult children talk to their parents about scams. The apparatus of prevention was aimed at the group everyone expected to be vulnerable. Young people received a different message: that they understood technology and the obvious scams were beneath them.

Tinder's own UK research makes the same point from the platform side. When the app launched its mandatory Face Check verification for new UK users in March 2026, it cited research that 63 per cent of 18 to 24 year olds said scams were harder to spot than ever. That is not a generation that thinks it is invincible. It is a generation that knows the ground has moved and has not been given better tools than a reverse image search and an instinct.

The Yin and Wen model also clarifies why “just be more careful” is weak advice. Verification is costly for an individual in a way it is not for a platform. A person on a dating app has almost no way to check whether a face has been synthetically generated, whether the same face appears on forty other accounts, whether the account was created yesterday from a device in a scam compound, or whether the text of the messages matches a script used on thousands of other victims. A video call can now be deepfaked; a reverse image search is defeated by a synthetic face. For an individual, the cost of meaningful verification is high and rising. For a platform, much of it is a database query.

Who Pays When the Check Is Missed

Britain already has an answer to the question of who pays when fraud succeeds, and it is instructive, because it does not include the platforms.

Since 7 October 2024, the Payment Systems Regulator has required banks and payment firms to reimburse victims of authorised push payment fraud, including romance scams, up to £85,000 per claim, with the cost split between the sending and receiving firms. Its dashboard shows around £316 million reimbursed in the eighteen months to March 2026, including £72.6 million in the first quarter of 2026 alone. The scheme covers Faster Payments and CHAPS; card payments, international transfers and other rails fall outside it, which matters because many pig-butchering scams move money onwards into cryptocurrency. Firms can refuse reimbursement where a customer was grossly negligent, a deliberately high bar, and only around 2 per cent of claims in the first quarter of 2026 were rejected on grounds of insufficient caution. The PSR itself is being folded into the Financial Conduct Authority.

The logic is that banks sit where money leaves the account, so bearing the cost gives them every incentive to detect fraud. But the transfer is the last moment in a romance scam, not the first. By the time money moves, the relationship has often run for weeks or months on a platform that took no financial share of the harm. In the language of the arXiv paper, the reimbursement regime assigns the cost of fraud to the one actor who can only verify at the very end, while the actors who host the fabrication and the targeting face no equivalent bill.

The platforms are not unregulated. Under the Online Safety Act, Ofcom's illegal content codes came into force on 17 March 2025, requiring services in scope to assess and mitigate the risk of illegal harms, fraud included. Ofcom's register of risks specifically identified romance fraud as one of the key risks associated with dating services. And on 10 July 2026, Ofcom opened a consultation on draft Fraudulent Advertising Codes of Practice, setting out nearly 40 measures for the largest platforms, including banning bad actors and blocking them from re-registering, with fines of up to £18 million or 10 per cent of global revenue for non-compliance once the codes are approved and in force. The consultation closes on 2 October 2026. But the advertising codes, as their name makes clear, cover paid-for advertising, not the user-generated profiles on which romance fraud actually runs.

The case for leaning harder on platforms is sharpened by what has emerged about their incentives. In November 2025, Reuters reported on internal Meta documents suggesting that the company had projected around 10 per cent of its 2024 revenue, roughly $16 billion, would come from advertising for scams and banned goods, and that its platforms showed users an estimated 15 billion higher-risk scam adverts a day. According to Reuters, Meta's own analysis found that universal advertiser verification would reduce fake adverts significantly but would cost revenue. Meta's spokesperson Andy Stone told Reuters the documents presented a “selective view”, that the 10 per cent estimate was rough and overly inclusive, and that the company had reduced user reports of scam adverts worldwide. The dispute concerns advertising rather than dating profiles, but it illustrates the structural problem: verification costs platforms money, while fraud costs them, at worst, reputation.

The dating industry has its own history here. In August 2025, Match Group agreed to pay $14 million to settle Federal Trade Commission charges relating to deceptive advertising, cancellation and billing practices. The FTC had alleged that Match sent non-subscribers notifications of interest that were in many cases from fraudulent accounts, without disclosing that fact, to induce them to subscribe. The following month, US Senators Marsha Blackburn and Maggie Hassan wrote to Match Group demanding answers on romance scams across its apps, which include Tinder, Hinge and OkCupid. Senator Hassan said that “Romance scams are robbing Americans of millions of dollars every year, and taking a devastating emotional toll in the process.”

What Platforms Could Verify That Users Cannot

To their credit, the major dating companies have started moving the burden in the right direction, and the early results support the argument that platforms, not users, are where verification is cheapest.

Tinder's Face Check asks new users to record a short video selfie during sign-up. The system checks that the face is live and matches the profile photographs, and it can detect the same face appearing across multiple accounts, which is exactly the signature of an operation running many synthetic suitors from one room. According to Tinder, the video itself is deleted after review, while an encrypted face map is retained to verify new photos and catch duplicates. Tinder said that in markets where Face Check is live it had seen a decrease of more than 60 per cent in exposure to potential bad actors and of more than 40 per cent in reports of bad actors. Yoel Roth, senior vice president of trust and safety at Match Group, called it “the most measurably impactful safety feature I've seen in my career.”

Bumble took a different route. In February 2024 it launched Deception Detector, a machine-learning system designed to identify spam, scam and fake profiles, and said that in testing it had automatically blocked 95 per cent of such accounts, with member reports of spam, scams and fake profiles falling by 45 per cent in the first two months. Bumble has also let users report profiles they believe use AI-generated photos or video.

These are company-reported figures, not independent audits. But they point to the central truth: signals that are invisible to an individual user are visible to the platform. A platform can see that an account was created minutes ago, that its device has hosted dozens of other accounts, that its photos may carry the statistical fingerprints of image generators, that its opening messages match a script sent to thousands of people, that it steers every conversation off the app to WhatsApp or Telegram within a day, or that it begins mentioning cryptocurrency platforms. A person on the other end of the chat sees only a charming stranger.

This is the Yin and Wen model turned into product design. If the danger is a protection gap in which verification lags fabrication, the fix is to make verification happen before trust accumulates rather than after money moves. Checking at sign-up, with a liveness test and duplicate detection, is verification at the cheapest possible moment: before the first message, before the grooming period, before the near-win. Checking at the bank transfer is verification at the most expensive moment, when the victim has been emotionally invested for weeks and will often override warnings to protect a relationship they believe is real.

The Limits of Putting It All on the Apps

None of this means the burden should move wholesale to platforms, and there are real costs to pretend away.

Biometric verification raises serious privacy questions. Storing face maps of millions of daters creates a target for attackers and a tool that could be misused. It also risks excluding people who cannot or will not submit their faces, including those with good reasons for caution, such as survivors of abuse or people whose sexuality could endanger them if exposed. Liveness checks themselves are a moving target: the same deepfake tools that fool a human on a video call are being aimed at the automated systems designed to catch them. And verification of identity is not verification of intent. A real, verified person can still be a fraudster, and the compounds UNODC describes are staffed by real people.

There is also the problem of where the conversation goes. Many romance scams begin on a dating app and move quickly to encrypted messaging, where the dating platform has no visibility. A fraudster verified out of one app will try another, and the arXiv model's logic applies here too: deception flows to the least vigilant venue. Uneven verification simply relocates the fraud, which is why cooperation across platforms matters as much as any one company's tools.

And individual judgement still matters. Lloyds' advice, to be wary of intense early declarations of love, to protect financial information, to resist urgency and guilt, and to talk to someone before sending money, remains good. The point is not that users should stop thinking. It is that users cannot be the primary verification layer for a threat that is industrial, automated and increasingly synthetic. Asking a 22-year-old to detect a real-time deepfake produced in a scam compound is like asking pedestrians to inspect the brakes on passing cars.

Keeping the Default Without Paying For It

The deepest cost of the synthetic suitor is not the money, large as the sums are. It is what the threat does to the ordinary human habit of believing people. If the only defence against fabricated romance is for everyone to trust less, then the fraudsters have taken something from people who were never scammed at all: the ease of meeting a stranger and assuming the best.

The alternative is to keep the truth default and move verification underneath it, into infrastructure, where it is cheap and systematic. Nobody inspects the wiring before switching on a light, because building regulations put the checking elsewhere. The benefit of the doubt survives there because someone with the tools and the responsibility has already done the doubting.

For online dating, that means treating verification at sign-up, duplicate-face detection, scam-script detection and friction when conversations move off-platform as baseline duties rather than optional features. It means Ofcom treating romance fraud on dating services as seriously as it is now treating scam adverts on social media. And it means asking, as British policy has not yet seriously asked, why the cost of romance fraud sits with banks at the final transfer, when the fabrication, targeting and grooming happen on platforms that currently pay nothing when it succeeds.

The Yin and Wen paper ends on a note of reassurance: trust arbitrage is self-limiting, because fraud eventually triggers the checking that stops it. In their model, the protection gap closes. The Lloyds data suggests it is still wide open, and the people falling into it are younger than anyone expected. The question is whether it closes because platforms verify earlier, or because a whole generation learns, one scam at a time, that kindness from a stranger is a warning sign. The first option costs companies money. The second costs everyone something harder to replace.

References and Sources

  1. Shaw, Vicky. “Romance scam reports have jumped with a surge in Gen Z victims, Lloyds figures.” PA Media, published via the Evening Standard and AOL UK, 25 September 2026. Reporting Lloyds Banking Group data for 1 September 2025 to 31 August 2026, including quotes from Liz Ziegler, Fraud Prevention Director, Lloyds, and Patrick Hurley, Ombudsman Director, Financial Ombudsman Service. https://www.aol.co.uk/articles/romance-scam-reports-jumped-surge-230100000.html
  2. The Independent. “Romance scams in UK rise by a quarter with huge surge in Gen Z victims.” September 2026, via inkl. https://www.inkl.com/news/romance-scams-in-uk-rise-by-a-quarter-with-huge-surge-in-gen-z-victims
  3. Foreign Policy Journal. “Romance Scam Reports Surge 24% At Lloyds (LON: LLOY) As Gen Z Victims Spike By 59%.” 25 September 2026. https://www.foreignpolicyjournal.com/2026/09/25/romance-scam-reports-surge-24-at-lloyds-lon-lloy-as-gen-z-victims-spike-by-59/
  4. Yin, Xieyu and Fenghua Wen. “When Trust Attracts Fraud: AI and Trust Arbitrage.” Business School, Central South University. arXiv:2609.27404, submitted 23 September 2026; journal reference Economics Letters, DOI 10.1016/j.econlet.2026.113234. https://arxiv.org/abs/2609.27404
  5. UK Finance. Annual Fraud Report 2026 and press release “Fraud remains a national security threat as criminals steal almost £1.3 billion.” June 2026. https://www.ukfinance.org.uk/policy-and-guidance/reports/annual-fraud-report-2026
  6. UK Finance and Take Five to Stop Fraud. “Gen Z more likely to be tricked by criminals and fall for impersonation scams.” 7 September 2023; OnePoll survey of 2,000 UK adults, 18 to 22 August 2023. https://www.takefive-stopfraud.org.uk/news/gen-z-more-likely-to-be-tricked-by-criminals-and-fall-for-impersonation-scams/
  7. Federal Trade Commission. “New FTC Data Show People Have Lost Billions to Social Media Scams.” Press release and Data Spotlight, 27 April 2026. https://www.ftc.gov/news-events/news/press-releases/2026/04/new-ftc-data-show-people-have-lost-billions-social-media-scams
  8. Federal Trade Commission. “Consumer Sentinel Network Data Book 2024.” March 2025, covering 2024 reports; see “Reported Fraud Losses by Age”. https://www.ftc.gov/system/files/ftc_gov/pdf/csn-annual-data-book-2024.pdf
  9. Levine, Timothy R. “Truth-Default Theory (TDT): A Theory of Human Deception and Deception Detection.” Journal of Language and Social Psychology, 2014. SAGE Publications. DOI 10.1177/0261927X14535916.
  10. Whitty, Monica T. “The Scammers Persuasive Techniques Model: Development of a Stage Model to Explain the Online Dating Romance Scam.” The British Journal of Criminology, volume 53, issue 4, July 2013, pages 665 to 684.
  11. Starling Bank. “AI voice cloning scams could catch millions out.” Research conducted with Mortar Research among more than 3,000 UK adults, 18 September 2024; see also CNN Business, “AI voice-cloning scams could target millions of people, Starling Bank warns”, 18 September 2024. https://www.starlingbank.com/news/starling-bank-launches-safe-phrases-campaign/
  12. CNN. “Deepfake romance scam raked in $46 million from men across Asia, police say.” 15 October 2024; and The Standard (Hong Kong), “Police arrest 27 for deepfake love scams totaling $360m, seizes scam-training documents”, October 2024. https://www.cnn.com/2024/10/15/asia/hong-kong-deepfake-romance-scam-intl-hnk
  13. WIRED. Reporting on Yahoo Boys' use of real-time face-swapping in romance scams, drawing on material gathered by David Maimon, Georgia State University and SentiLink, 2024; see also AI Incident Database, Incident 911. https://incidentdatabase.ai/cite/911/
  14. United Nations Office on Drugs and Crime. “An Interconnected Criminal Ecosystem: Transnational Organized Crime Threat Assessment for South-East Asia 2026.” Launched in Bangkok, 21 July 2026; see also OCCRP, “UN Report Details Southeast Asia's Interconnected Criminal Economy”, 21 July 2026. https://www.occrp.org/en/news/un-report-details-southeast-asias-interconnected-criminal-economy
  15. Meta and WhatsApp. Announcement of the removal of 6.8 million WhatsApp accounts linked to criminal scam centres in the first half of 2025, August 2025; as reported by the Associated Press via US News, “WhatsApp Takes Down 6.8 Million Accounts Linked to Criminal Scam Centers, Meta Says”, 6 August 2025. https://www.usnews.com/news/business/articles/2025-08-06/whatsapp-takes-down-6-8-million-accounts-linked-to-criminal-scam-centers-meta-says
  16. Horwitz, Jeff. “Meta is earning a fortune on a deluge of fraudulent ads, documents show.” Reuters, November 2025, including the response of Meta spokesperson Andy Stone; republished by Editor and Publisher. https://www.editorandpublisher.com/stories/meta-is-earning-a-fortune-on-a-deluge-of-fraudulent-ads-documents-show,258660
  17. Payment Systems Regulator. “APP fraud reimbursement protections” and “APP scams reimbursement dashboard for Q1 2026.” https://www.psr.org.uk/information-for-consumers/app-scams-reimbursement-dashboard/
  18. Payment Systems Regulator. PS25/5, Consolidated policy statement, APP scams reimbursement requirement, May 2025; and Consumer Standard of Caution guidance, December 2023. https://www.psr.org.uk/media/rhelv4op/ps25-5-app-scams-reimbursement-consolidated-policy-statement-may-2025.pdf
  19. HM Treasury. “A Streamlined Approach to Payment Systems Regulation: Consultation response.” GOV.UK, 2026, confirming the consolidation of the Payment Systems Regulator into the Financial Conduct Authority. https://www.gov.uk/government/consultations/a-streamlined-approach-to-payment-systems-regulation-consultation/outcome/a-streamlined-approach-to-payment-systems-regulation-consultation-response
  20. Ofcom. Illegal content Codes of Practice for user-to-user services, issued 24 February 2025, in force 17 March 2025; and “Protecting people from illegal harms online: Register of Risks”, 16 December 2024. https://www.ofcom.org.uk/siteassets/resources/documents/online-safety/information-for-industry/illegal-harms/register-of-risks.pdf
  21. Ofcom. “Consultation: Fraudulent Advertising Codes of Practice.” Published 10 July 2026, closing 2 October 2026. https://www.ofcom.org.uk/online-safety/online-fraud/consultation-fraudulent-advertising-code-of-practice
  22. Tinder. “Tinder Launches Face Check in the UK, Setting a New Standard for Dating Safety.” Tinder UK Press Room, March 2026, including quotes from Yoel Roth, Match Group, and Mark Tierney, Stop Scams UK. https://uk.tinderpressroom.com/news?item=122625
  23. Bumble Inc. “Bumble Inc. Launches Deception Detector: An AI-Powered Shield Against Spam, Scam and Fake Profiles.” Business Wire, 6 February 2024. https://www.businesswire.com/news/home/20240206911584/en/
  24. Federal Trade Commission. “Match Group Agrees to Pay $14 Million, Permanently Stop Deceptive Advertising, Cancellation, and Billing Practices to Resolve FTC Charges.” Press release, 12 August 2025. https://www.ftc.gov/news-events/news/press-releases/2025/08/match-group-agrees-pay-14-million-permanently-stop-deceptive-advertising-cancellation-billing
  25. Office of Senator Marsha Blackburn. “Senators Blackburn, Hassan Press Match Group for Answers on Romance Scams.” 24 September 2025. https://www.blackburn.senate.gov/index.php/2025/9/crime/technology/senators-blackburn-hassan-press-match-group-for-answers-on-romance-scams

Tim Green

Tim Green UK-based Systems Theorist & Independent Technology Writer

Tim explores the intersections of artificial intelligence, decentralised cognition, and posthuman ethics. His work, published at smarterarticles.co.uk, challenges dominant narratives of technological progress while proposing interdisciplinary frameworks for collective intelligence and digital stewardship.

His writing has been featured on Ground News and shared by independent researchers across both academic and technological communities.

ORCID: 0009-0002-0156-9795 Email: tim@smarterarticles.co.uk

Listen to the free weekly SmarterArticles Podcast

Discuss...