.png)
In 2025, Americans lost $11.37 billion to cryptocurrency scams, and a growing share of that money left their accounts because of a face or a voice that was not real. Deepfakes have become the connective tissue of two of the most damaging fraud categories online: romance scams and investment fraud. They let a stranger look like a devoted partner on a video call, and they let a fabricated Elon Musk promise life-changing returns in a slick advertisement.
This guide explains how deepfakes fuel romance and investment fraud: where synthetic media plugs into each scam, the specific techniques criminals use, why the schemes are so effective, and what actually works to detect and stop them. It is written for trust and safety teams, fraud and compliance staff, and anyone who wants to understand the threat past the headlines.
The two categories increasingly merge into one. In the model known as romance baiting, financial grooming, or pig butchering, a scammer spends weeks building a relationship before steering the victim into a fake crypto investment. Global losses from that single pattern have topped $75 billion since 2020, according to University of Texas research, and deepfakes are what make the trust believable.
- Deepfakes now sit at the center of romance baiting (also called pig butchering or financial grooming), which blends romance and investment fraud.
- Scammers use real-time face-swap on live video calls, so a video chat is no longer proof the person is real.
- The most impersonated figure in celebrity investment-scam deepfakes is Elon Musk, often pushing a fake platform called Quantum AI (Sensity).
- One 82-year-old US retiree lost $690,000, and an Ontario victim lost $1.7 million, to Musk-deepfake crypto scams.
- Americans lost $11.37 billion to crypto scams in 2025, up 22% year over year, with victims 60 and older bearing $4.35 billion (FBI IC3).
- The FBI attributed $893 million in 2025 losses specifically to AI-powered scams, including deepfake endorsements and pig butchering.
- Global pig-butchering losses have topped $75 billion since 2020 (University of Texas research).
- Defense means treating video and ads as unverified: verify out-of-band, never invest on the strength of an endorsement, and deploy detection at the platform level.
What Are Deepfake-Powered Romance and Investment Scams?
A romance scam uses a fabricated identity to build emotional intimacy, then converts that trust into money. An investment scam persuades a victim to put funds into a fake opportunity, usually cryptocurrency. The dominant modern hybrid, which INTERPOL prefers to call romance baiting and AARP calls financial grooming, fuses the two: the relationship is the setup, and the fake investment is the payoff.
Deepfakes are the technology that makes the fabricated identity hold up under scrutiny. A generated face can front a dating profile, a real-time face-swap can survive a video call, a cloned voice can carry a phone conversation, and a synthetic video of a trusted public figure can vouch for a bogus platform. In every case the deepfake attacks the same thing: the victim's ability to confirm that what they are seeing and hearing is real. For a closely related use of the same techniques against identity checks, see our guide to how deepfakes bypass KYC.
How Deepfakes Enter the Scam
Synthetic media plugs into these scams at four distinct points, and most large operations use more than one.
- The persona. An AI-generated face or a stolen-and-altered photo creates a profile that reverse image search will not immediately unmask.
- The proof. When a victim asks to video chat to confirm the person is real, a real-time face-swap lets the scammer appear as someone else entirely, in sync, on a live call.
- The voice. A cloned voice, built from seconds of audio, handles phone calls and voice notes so the persona sounds consistent across channels.
- The endorsement. For investment fraud, a deepfake of a celebrity or public official provides the authority that convinces a stranger to deposit money.
Understanding these four entry points matters because each calls for a different defense, and because a scam that fails at one can still succeed through another.
Deepfakes in Romance Fraud
Synthetic Profiles and Personas
Every romance scam begins with a believable identity. Generative tools let a scammer produce a face that belongs to no one, sidestepping the old defense of reverse-image-searching a stolen photo. Combined with large language models that write fluent, emotionally attuned messages in any language, a single operator can now run many relationships at once. Criminal organizations use automated sentiment analysis to adjust tone based on how each victim responds, which is how compounds in Southeast Asia manage thousands of simultaneous conversations.
Real-Time Deepfake Video Calls
For years, the standard safety advice was to insist on a video call, on the theory that a scammer could not fake a live face. That advice is now obsolete. Using face-swap software fed through a virtual camera, a scammer can appear on a live call as a completely different person, with the AI mirroring their expressions and mouth movements in real time. The video chat that used to be the ultimate proof of authenticity is now just another prop.
The scale is real. In October 2024, Hong Kong police broke up a syndicate of 27 people that used AI face-swapping and voice-changing tools to build fake personas on dating platforms, then lured victims into fabricated cryptocurrency investments, with victims convinced by real-time deepfake video calls losing millions collectively. Individual cases are devastating: a Connecticut woman lost nearly $1 million after a scammer sustained a relationship with these tactics over many months, and the FBI's Norfolk field office confirmed in early 2026 that criminals now routinely add AI-generated voice messages to make the schemes more believable.
Deepfakes in Investment Fraud
Celebrity Endorsement Deepfakes
The investment side runs on borrowed authority. Scammers generate videos of famous, trusted figures appearing to endorse a fake platform, then pay to promote them as ads. According to a 2024 report from the detection firm Sensity, Elon Musk is the most impersonated figure in these scams, frequently pushing a fabricated crypto platform called Quantum AI, sometimes alongside faked testimonials from other celebrities. The financial damage is severe. Steve Beauchamp, an 82-year-old retiree, drained his retirement fund and invested $690,000 in a deepfake Musk scheme, and an Ontario victim lost $1.7 million to a similar AI Musk video.
The cast of impersonated figures is long and local to each market. In the UK, deepfake ads have featured Martin Lewis, Keir Starmer, Piers Morgan, and Richard Branson; in Canada, Prime Minister Mark Carney and celebrities such as Michael Bublé. A woman in Salisbury, England lost around 20,000 pounds after seeing a deepfake ad in which the UK prime minister appeared to promote a crypto scheme. Celebrity deepfakes were the single most common type of scam advert reported to the UK Advertising Standards Authority in 2024.
Fake Live Streams and Spoofed News
Two variations add urgency and false credibility. In the first, scammers upload a prerecorded deepfake to a streaming platform under a "live" label, showing a figure like Musk promising to personally double any cryptocurrency sent to a wallet, drawing large audiences before the video is removed. In the second, ads link to fabricated news articles that copy the look of outlets like the BBC, complete with a deepfake interview, to make the fake platform look independently reported. The distribution problem is enormous: internal Meta documents cited in 2026 lawsuits projected that roughly $16 billion, about 10% of the company's 2024 revenue, came from fraudulent advertising, a category that includes deepfake celebrity endorsements.
Why These Scams Work
Deepfakes succeed because they target instincts that operate below conscious scrutiny. A romance scam exploits attachment: once a victim believes they are in a relationship, a live video call that appears to show their partner shuts down doubt rather than raising it. An investment scam exploits authority bias: a familiar, trusted face appearing to endorse a platform borrows credibility that the platform never earned. Both add urgency, whether a partner's sudden crisis or a closing investment window, because urgency is what prevents verification.
The problem is compounded by the fact that people cannot reliably tell the difference anymore. As synthetic media improves, the visual and audio tells that used to give a fake away are disappearing, and any defense that depends on a human noticing something looks off is already losing. Seeing is no longer believing, and scams are built precisely around the assumption that it still is.
Common Mistakes and Missed Red Flags
Treating a video call as proof. A live video is no longer evidence that a person is who they claim to be, because real-time face-swaps are now widely available.
Trusting a celebrity endorsement in an ad. No legitimate investment is sold through a social media video of a celebrity promising guaranteed or doubled returns. Authority in an ad is exactly what the scam manufactures.
Following the move to a private channel. A push to shift the conversation to WhatsApp, Telegram, or Signal is a common early step, because those channels are harder for platforms to monitor.
Mistaking small early returns for legitimacy. Fake platforms often allow a small initial withdrawal to build confidence before the larger deposits that are never returned.
Ignoring love bombing. Rapid, intense affection within days, with pet names and talk of a shared future, is a well-tested manipulation script, not a fast connection.
How to Detect and Defend Against Deepfake Fraud
Because the media itself can no longer be trusted, effective defense pairs technical detection with verification habits that do not depend on the audio or video looking right.
At the individual level, several tells can still surface a real-time deepfake on a live call, though none is conclusive on its own. Watch for lip movements that lag or look rubbery on open vowels, blinking that is too rare or too frequent, blurring or artifacts around the hairline when the person turns, and lighting on the face that does not match the room. Asking the caller to move naturally or turn their head sideways can stress a face-swap that only renders a front-facing view. The stronger defense is behavioral: verify identity out-of-band, never send money or invest on the basis of a call or an advertisement alone, and confirm any celebrity endorsement through official channels rather than the ad itself.
At the platform level, where dating apps, social networks, and exchanges bear the burden, detection has to be automated and applied at scale. That means screening profile images and ad creative for synthetic-media signatures, analyzing live video for the artifacts a real-time swap leaves behind, and flagging cloned audio in calls. This is where dedicated detection belongs. DuckDuckGoose's DeepDetector analyzes images and video for the inconsistencies that betray synthetic faces, so a platform can catch a deepfake persona in a profile, a face-swap in a live call, or a fabricated celebrity in an ad before it reaches a user. Mordor Intelligence names DuckDuckGoose AI among the major companies in the fake image detection market.
Frequently Asked Questions
How do deepfakes fuel romance and investment fraud?
Deepfakes let a scammer maintain a believable false identity and borrow authority. A generated face fronts a dating profile, a real-time face-swap survives a video call, a cloned voice handles phone conversations, and a synthetic video of a celebrity endorses a fake investment platform. Each removes a way the victim might otherwise verify that the person or opportunity is real.
What is pig butchering, and how do deepfakes relate to it?
Pig butchering, also called romance baiting or financial grooming, is a scam that blends romance and investment fraud: the criminal builds a relationship over weeks, then lures the victim into a fake crypto investment. Deepfakes make the fabricated partner convincing, including through real-time video calls that appear to confirm the person is real.
Can a video call prove someone is real?
No longer. Face-swap software fed through a virtual camera lets a scammer appear as a different person on a live call, mirroring expressions and speech in real time. A video call should be treated as one signal, not proof, and should be backed by independent verification.
Why is Elon Musk used in so many deepfake scams?
He is the most impersonated figure in investment-scam deepfakes, according to Sensity, largely because of his association with wealth and technology and the vast amount of public video of him, which makes convincing clones easier to build. Scammers commonly pair his likeness with a fake platform such as Quantum AI.
What are the warning signs of a deepfake in a live video call?
Look for lip movements that lag or look rubbery on open vowels, unnatural blinking, blurring or distortion around the hairline when the person turns, and facial lighting that does not match the surroundings. These signs are helpful but not definitive, so pair them with out-of-band verification.
How can dating and social platforms defend against deepfake scams?
By automating detection at scale: screening profile photos and advertisement imagery for synthetic-media signatures, analyzing live video for face-swap artifacts, and flagging cloned audio. Platform-level facial verification and detection tools reduce exposure before a scammer ever reaches a user.
How much money do these scams cause?
Losses are large and rising. Americans lost $11.37 billion to crypto scams in 2025, up 22% year over year, and the FBI attributed $893 million specifically to AI-powered scams. Global pig-butchering losses have exceeded $75 billion since 2020.
How can I verify a celebrity investment endorsement?
Assume it is fake until proven otherwise. Legitimate opportunities are not sold through social media videos promising guaranteed or doubled returns. Check the public figure's official website and mainstream news coverage, and consult a licensed financial professional before investing.
About the author
