Skip to main content
Back to Blog

WellWired Journal

AI Romance Scams UK: How to Spot a Fake

UK romance scam victims lost £102 million in 2025, with over half aged 55 or older. Plain-English guide to spotting AI fakes and reporting in the UK.

9 July 202614 min readBy Arthur Turing
AI Romance Scams UK: How to Spot a Fake

Quick summary: UK romance scam victims lost £102 million in 2025. Over half were aged 55 or older. Scammers now build fake relationships using AI-generated profile photos, chatbot conversations, cloned voices, and deepfake video. Seven warning signs still work: moving off the platform quickly, never agreeing to meet in person, glitchy or cancelled video calls, a profile that looks too perfect, early declarations of love, requests for money, and demands for gift cards or cryptocurrency. If you think you are being scammed, stop contact and call your bank immediately. Report to Report Fraud on 0300 123 2040.

A photo arrives through a dating app or social media. A widower. Works as a civil engineer on an oil rig. Thoughtful eyes, silver temples, a warm smile. His first message says he stumbled across your profile and felt he had to say hello. Over the weeks that follow, the messages come every morning. He asks about your day, your family, the things you enjoyed when you were younger. He remembers every detail. He feels like the relationship you had stopped hoping for.

That photograph was created by an AI. The messages are produced by a chatbot, or written from a script by a person working in a fraud operation abroad. The emotional connection you feel is real. The person on the other end is not.

This is what AI romance scams look like in 2026. Report Fraud data for 2025 shows 10,784 romance scam cases in the UK, a 29% year-on-year increase, with total losses of £102 million. That works out to roughly £280,000 stolen every single day (calculated from the annual total). Our wider guide to AI phone scams in the UK covers related tactics criminals use over the phone.

What is an AI romance scam?

A romance scam is when someone pretends to develop a romantic relationship with you in order to steal money or personal information. The crime is not new. What changed is the technology criminals use to run it.

Until a few years ago, a romance scam required a person to maintain the deception manually, which limited how many victims a criminal could target at once. AI has removed that constraint. Criminals can now generate a convincing fake identity within minutes. AI writing tools handle the daily conversation. Software generates profile photos of people who do not exist. Entire operations run at a scale that was simply impossible before.

The Alan Turing Institute's Centre for Emerging Technology and Security documented how AI is being used to automate romance fraud from end to end: synthetic personas, scripted conversation, fabricated video content, and coordinated financial extraction. These are organised operations, not opportunistic crimes.

Why are older adults most at risk?

The data points in the same direction across every report: older adults are both more often targeted and more likely to lose larger amounts. Report Fraud figures for 2025 show that adults aged 55 to 74 accounted for almost half of all romance scam losses in the UK, and that over-55s were victims in 58% of all romance fraud cases. TSB fraud data cited in the same report found the 65 to 74 age group the most frequently targeted, accounting for 23% of all reported cases. The average per-victim loss was £9,500, with the highest single case in 2025 reaching £1 million.

This targeting is deliberate. Retirement savings are more likely to be accessible. People living alone are more likely to welcome a warm, attentive contact. And anyone who did not grow up with social media may not know how quickly a convincing profile can be put together from nothing. Being deceived by one of these scams says nothing about your intelligence. It says something about how good the technology has become.

What AI tools do romance scammers use?

Understanding the tools makes the warning signs easier to spot.

AI-generated profile photos

Photo-generation software can produce a realistic image of a person who does not exist within seconds. The photos typically show someone attractive and professional-looking, often claiming to work abroad in the military or a professional field like engineering. The image looks like a genuine personal photo, but no such person exists. Our guide to spotting deepfake images and videos covers the visual tells: look for hands with the wrong number of fingers, backgrounds that blur where they should be crisp, or ears that look subtly mismatched on each side. A reverse image search using Google Images can sometimes reveal the same photo appearing on multiple profiles under different names.

AI chatbots handling conversation

AI writing tools can maintain a persuasive conversation across weeks or months. The chatbot asks questions that feel personal, remembers your answers, and sends messages at predictable times to simulate a daily routine. Signs to watch for: replies that miss the emotional weight of what you said, repeat phrases from earlier, or feel oddly polished. Ask an unexpected, specific question and see how naturally the answer comes back.

Voice cloning

When a voice call does happen, it may not be the person you think. Voice-cloning software can replicate a person's voice from a short audio sample. If the scammer's profile links to any video or audio content, that voice can be copied and used in calls. Our guide to AI voice scams in the UK explains how this works and why even a familiar-sounding voice is no longer a reliable indicator of trust.

Deepfake video

Some operations use deepfake video to add another layer of apparent authenticity. Short clips can be generated by AI, or live video calls can be filtered in real time to overlay a false face. Ask the person to do something specific and spontaneous during a call: wave both hands in front of their face, or put a hand over the camera and remove it slowly. AI deepfake tools often struggle with sudden hand movements and produce momentary glitches at the face edges. Research from Barclays in early 2026 found 44% of UK adults said they would not be confident spotting voice cloning technology or a deepfake in a video call.

Seven warning signs that still work

The technology may have changed. These signals have not.

  1. They move off the platform quickly. A scammer suggests switching to WhatsApp or another private messaging app almost immediately. Dating apps and social platforms have reporting tools and moderation. Private messaging does not. The suggestion to move elsewhere very early is a flag worth taking seriously.
  2. They never agree to meet in person. There is always a reason. A work project overseas. A family crisis. A last-minute emergency. Meetings are always almost about to happen, and then something gets in the way. This pattern, repeated over weeks or months, is one of the most reliable signals of all.
  3. Video calls are cancelled or technically strange. The camera freezes. The connection drops. The call is rescheduled and then does not happen. Or the video call does occur, but something is slightly off: the face moves fractionally ahead of or behind the audio, or does not track naturally when they turn their head.
  4. The profile looks too perfect. Attractive, successful, recently widowed, grown-up children, a job that keeps them away from the UK for long stretches. These are the stock elements of a constructed persona. No real person ticks every box in quite the same way.
  5. They declare love very quickly. Deep feelings expressed within days or a few weeks are a warning sign, not evidence of chemistry. Real relationships take longer to build. Fraudsters accelerate emotional connection because it makes the requests that follow harder to resist.
  6. They ask for money, or pitch an investment. The request may arrive gradually: an emergency, a medical crisis, money temporarily inaccessible and needing a short-term bridge. Or it may be framed as an investment opportunity they want to share. However it arrives, money moving from you to them is the goal of the entire exercise.
  7. They cannot receive money by normal bank transfer. Scammers avoid traceable payment methods. They prefer transfers to unfamiliar accounts, gift cards (asking for the card codes), wire transfers, or cryptocurrency. A request for gift card codes is a confirmed scam. No legitimate contact ever needs gift cards sent to them.

What should you do if you think you are being scammed?

It can be hard to accept that a relationship you have invested in may not be real. That difficulty is part of what the scam depends on. Try to treat it as a practical problem rather than a question about your judgement.

  1. Stop contact. Do not reply. Do not explain your decision. Do not try to confront them or ask for a confession. Just stop responding. In most cases, cutting contact ends the pressure immediately.
  2. Do not send any more money. Even if you receive messages that seem distressing or urgent. Emotional pressure is the main tool used at this point in the process, and responding to it almost always results in further loss.
  3. Tell someone you trust. A family member, a close friend, or your GP. Telling someone helps you stay firm about not sending money. It also makes the practical steps below easier to take.
  4. Reverse image search their profile photo. On Google Images, click the camera icon and upload the photo. On a phone, hold your finger on the image and choose the search option. If the same photo appears on multiple profiles under different names, the profile is confirmed as fake.
  5. Contact your bank. If you have already sent money, call your bank immediately on the number on the back of your card. The sooner you report it, the more options are available to you.

What if you have already sent money?

Under rules introduced by the Payment Systems Regulator in October 2024, UK banks must reimburse victims of authorised push payment fraud. This covers situations where you were deceived into transferring money yourself. Banks can deduct up to £100 from a standard reimbursement, but the obligation to pay back the rest exists. The scheme covers losses up to a maximum of £85,000 per claim; losses above that threshold fall outside the mandatory scheme, though they should still be reported. PSR monitoring data from the scheme's first year showed the majority of APP fraud victims receiving at least partial reimbursement.

  1. Call your bank immediately on the number from the back of your card. Tell them what happened and when. Ask them to raise a fraud claim and trace the payment if possible.
  2. Keep all messages. Do not delete the conversation. Screenshots of every message and any profile details will support your fraud claim. Your bank will ask for evidence.
  3. Report to Report Fraud (see below). A crime reference number helps progress claims with most banks.
  4. Contact Citizens Advice or Victim Support if you need help working through the process, or are finding it hard to talk about.

If the transfer was made in cryptocurrency, the position is harder. Cryptocurrency payments are generally irreversible and fall outside the bank reimbursement scheme. Include the transfer in your Report Fraud submission regardless: the information contributes to investigations even when direct recovery is unlikely.

For more detail on how banks handle AI-related fraud and what you can safely share with digital tools, our guide to AI and online banking covers that in depth.

How do you report a romance scam in the UK?

Reporting matters even when money is gone. Every case filed adds to the intelligence picture that police use to track fraud networks.

Report Fraud is the national reporting service for fraud in England, Wales, and Northern Ireland. You can report online at reportfraud.police.uk or by calling 0300 123 2040 (Monday to Friday, 8am to 8pm). In Scotland, contact Police Scotland on 101.

Report the profile on the platform where you met. Dating apps and social media platforms have reporting tools. Each report helps prevent the same account from being used against someone else.

The Take Five to Stop Fraud campaign, backed by UK Finance and the Home Office, offers practical guidance at takefive-stopfraud.org.uk. Their core principle is simple: pause before acting, question whether the request makes sense, and do not share money or information in response to something you did not initiate.

For emotional support, Victim Support offers free, confidential help at 08 08 16 89 111. Feeling embarrassed or ashamed after a romance scam is common, and it is exactly what fraudsters rely on to stop people reporting. Police and bank fraud teams are well used to handling these cases with full discretion. The shame belongs with the people who ran the scam, not with the people they deceived.

If you want to build your confidence with AI and technology generally, the WellWired Academy covers AI for everyday life in plain English, with no technical knowledge assumed. And for a closer look at how criminals use AI-generated voices in phone calls, our guide to voice cloning in the UK explains what these calls sound like and what to do when you receive one.

FAQ

How can you tell if someone is using AI in an online relationship?

The clearest signs: they never agree to meet in person despite weeks of promises; video calls are cancelled or produce glitches; their profile photo returns other results in a reverse image search; their messages feel slightly scripted or miss the emotional weight of what you said; and at some stage they ask for money or propose an investment. No single sign is definitive, but several together are a strong signal.

Are romance scams always run by real people?

Not always. AI chatbots now handle much of the early conversation, particularly on dating apps and social platforms. A human operator may step in for voice or video calls, or when the financial request is made. In some operations, the entire relationship from first message to money request runs on automation. This is part of why these scams can now operate at such scale.

What does a romance scammer say to make you trust them?

Scammers use consistent patterns: early and intense expressions of connection; questions that feel genuinely interested in your life; a backstory designed to invite sympathy, such as being recently widowed or working abroad in a demanding job; and gradual escalation toward financial conversation. They are often very patient. Some spend months before making any request at all. The slow build is deliberate.

Can you get money back from a romance scam?

Possibly. Under rules introduced in October 2024, UK banks must reimburse most victims of authorised push payment fraud (bank transfers made while being deceived), up to a maximum of £85,000 per claim. Contact your bank straight away, keep all messages as evidence, and report to Report Fraud on 0300 123 2040. Cryptocurrency payments fall outside this scheme and are much harder to recover.

Is it embarrassing to report a romance scam?

Many people feel ashamed, and that feeling is exactly what scammers rely on to stop people reporting. These are sophisticated, well-funded operations using tools that fool people of all ages and backgrounds. No one should feel embarrassed about being targeted. Police and bank fraud teams treat romance fraud victims with full seriousness. Victim Support (08 08 16 89 111) offers confidential help for anyone who finds this hard to talk about.

Are romance scams only on dating apps?

No. Romance scams now frequently begin on Facebook, Instagram, and other social platforms, including fitness and hobby apps. The first contact may look like a random connection request or a message that seems to have been sent by mistake. Once contact is established, the conversation typically moves to a private messaging app with no fraud-reporting tools.

Ai Romance ScamsRomance Scams UkAi Dating ScamsHow To Spot A Romance ScamRomance Fraud Uk
Free AI Starter Kit

Start with one calm, practical guide.

A friendly 5-page guide to help you understand AI, know what to try first, and avoid the most common mistakes. You'll also get a weekly plain-English email. Unsubscribe anytime.

5-page beginner guideWritten for over 50sNo tech jargon

We'll send the guide straight away.

About the Author

Arthur Turing avatar
Arthur TuringCEO & Lead Writer

Arthur is WellWired's founder and lead writer.

Want to keep learning?

Explore more in-depth guides or start a structured learning path built for beginners.