An AI deepfake scam is a form of fraud in which criminals use artificial intelligence tools to generate fake audio, video, or images that convincingly impersonate a real person — and then use those fabrications to deceive victims into transferring money, providing sensitive information, or taking actions that benefit the fraudster. The word “deepfake” combines “deep learning” — the branch of artificial intelligence used to generate the content — with “fake.”

The AI deepfake scam takes several forms. Voice cloning creates a synthetic copy of a person’s voice that can say anything the fraudster types. Face-swap deepfakes place one person’s face onto another person’s body in a video. Full synthetic video generation creates entirely new video footage of a person who appears to be speaking in real time. Each of these technologies is now being weaponised in the AI deepfake scam against individuals, businesses, and institutions worldwide. The threat is no longer confined to high-value corporate targets — it now reaches ordinary families, small businesses, and retirees with equal effectiveness.

What makes the AI deepfake scam uniquely dangerous is its ability to destroy the very trust mechanisms people have traditionally relied upon to identify fraud. When a phone call sounds exactly like your daughter’s voice, when a video call shows your CEO’s face delivering urgent instructions, or when a social media video appears to show a trusted celebrity endorsing an investment platform, the normal warning signs that protect most people from fraud simply do not apply. The AI deepfake scam bypasses rational scepticism by exploiting the most fundamental human instinct — to trust what we see and hear with our own senses. The same technology powers the grandparent variant in our Amazon gift card scam guide and the celebrity-endorsement variant in our crypto investment scams guide.

Almost every AI deepfake scam follows the same five-stage pattern, from harvesting social media clips to the moment the money is transferred.

Step 1: Collecting Audio and Video Material

The AI deepfake scam begins with the criminal collecting publicly available audio or video material of their intended target. Social media platforms are the primary source — videos posted on Facebook, Instagram, TikTok, LinkedIn, and YouTube provide criminals with hours of audio and visual material from which to build convincing deepfakes. For corporate targets, earnings calls, conference presentations, and media interviews are harvested. For private individuals, voice messages, video calls, and personal social media posts are exploited. Modern AI voice cloning tools require as little as three to ten seconds of audio to create a convincing synthetic voice. Video deepfake tools require a collection of photographs or video frames of the target’s face. The abundance of material most people post publicly on social media means that for the majority of targets, the AI deepfake scam operator can build their fabrication tools quickly and at essentially no cost.

Step 2: Creating the Deepfake Content

Once sufficient source material has been collected, the criminal feeds it into an AI tool to create the fraudulent content. For voice cloning variants of the AI deepfake scam, the criminal types the script they want the cloned voice to deliver — the AI tool generates a synthetic audio file that sounds like the target saying those exact words. For video deepfakes, the criminal either swaps the target’s face onto existing video footage or generates new video in which the target appears to be speaking. The quality of AI deepfake scam content has improved dramatically. In a phone call or a compressed video call, voice clones and face-swapped videos are now virtually indistinguishable from genuine recordings for most people, particularly when the content is delivered with urgency that prevents the recipient from taking time to analyse what they are experiencing.

Step 3: Delivering the Fraudulent Contact

The AI deepfake scam is then deployed through whichever communication channel is most likely to succeed. Phone calls using cloned voices are the most common delivery method for personal versions — the caller ID is often spoofed to appear as the genuine contact. Video calls through WhatsApp, Zoom, or Teams are used for corporate and business deepfake fraud. Social media and YouTube are used to distribute celebrity endorsement deepfakes promoting fraudulent investment platforms. Email and messaging apps deliver written instructions accompanied by convincing AI-generated audio or video attachments. The delivery channel is chosen to match the impersonation — phone calls feel personal, video conferences feel professional, social media feels broadcast.

Step 4: Creating Urgency and Pressure

Like all effective fraud, the AI deepfake scam relies heavily on urgency and pressure to prevent the victim from pausing to verify what they are experiencing. A cloned voice call claiming a family member has been in an accident and urgently needs money. A deepfake video call from an apparent CEO demanding an immediate wire transfer before a deal closes. A fake celebrity investment video claiming a limited-time opportunity that will expire within hours. The manufactured urgency of the AI deepfake scam is designed to trigger an emotional response that overrides rational scepticism — the same isolation tactic documented in our imposter scam warning signs guide.

Step 5: Collecting the Money or Data

Once the victim is convinced they are communicating with a legitimate person or organisation, the AI deepfake scam operator directs them to transfer money, provide bank account details, share login credentials, or invest in a fraudulent platform. Payment methods requested in these scams are typically those that offer little recourse — bank transfer, cryptocurrency, or gift card codes. Once payment is made or credentials are shared, the fraudster disappears and the victim is left to discover that the entire interaction was fabricated.

The AI deepfake scam adapts to whichever trust relationship the criminal can exploit — family, employer, celebrity, romantic partner, or identity verification system. These are the five most reported variants.

The Parent Who Heard Their Child’s Voice

The AI deepfake scam strikes hardest when it weaponises a parent’s love. A woman in her sixties received a phone call from what appeared to be her adult son’s mobile number. The voice on the line was unmistakably her son’s — the same tone, the same accent, the same way of saying her name. He told her he had been in a car accident abroad, had been arrested, and needed $3,000 wired immediately to pay a lawyer before he could be released. He begged her not to call anyone else as it would make the legal situation worse. She was in the process of arranging the transfer when her daughter called on another line — she had just spoken to her brother, who was at home and completely unaware of what was happening. The AI deepfake scam had cloned her son’s voice from videos on his social media profile. Had her daughter not called at that moment, she would have lost $3,000 and the emotional experience of hearing what she believed was her son in distress would have been deeply traumatic regardless of whether the money was recovered.

The Finance Director Who Transferred $25 Million

The AI deepfake scam now reaches the highest levels of corporate finance. A finance professional at a multinational company in Hong Kong received an invitation to join a video conference call with several senior colleagues, including the company’s CFO, to discuss a confidential business transaction. Every participant on the call appeared exactly as expected — their faces, voices, and professional demeanour all appeared completely genuine. The CFO instructed the finance director to make a series of wire transfers totalling the equivalent of $25 million as part of the transaction. The finance director complied. It was only later, when the real CFO was contacted about the transaction, that the fraud was discovered. Every participant on that call had been a deepfake. This AI deepfake scam — executed with extraordinary sophistication — resulted in one of the largest single fraud losses ever attributed to deepfake technology.

The Retiree Who Invested in a Fake Platform

The AI deepfake scam reaches retirees through social media advertising. A retired teacher saw a video on social media in which a well-known business personality appeared to describe a new investment platform that had generated exceptional returns for early investors. The video was professional, articulate, and convincing — the business personality discussed specific investment strategies and urged viewers to act quickly to secure their position. She invested £15,000 in the platform. When she attempted to withdraw her funds several months later, she was told further fees were required. She paid these fees. The platform then became unresponsive. The video was an AI deepfake scam — the business personality had never made or endorsed the video, and the investment platform had been fraudulent from the start. Her retirement savings were gone.

Law enforcement and regulatory bodies worldwide are increasingly focused on the AI deepfake scam threat, though the pace of technological change presents significant enforcement challenges.

The Federal Bureau of Investigation in the United States has issued multiple public warnings about the AI deepfake scam, particularly voice cloning fraud targeting families and business email compromise using deepfake video. The FBI advises the public to establish family code words for emergencies and to verify any urgent financial request through a separately initiated call to a known number before taking action. Report at ic3.gov.

Action Fraud in the United Kingdom accepts reports of the AI deepfake scam through its online reporting tool at actionfraud.police.uk and by telephone on 0300 123 2040. The National Cyber Security Centre also maintains guidance on deepfake fraud and a suspicious content reporting service at ncsc.gov.uk.

The European Union’s law enforcement agency Europol has published threat assessments identifying the AI deepfake scam as among the most significant emerging crime threats and has called for coordinated international legislation to address the creation and distribution of malicious synthetic media.

Multiple governments are introducing or have introduced legislation specifically targeting malicious deepfakes. In the United States, several states have enacted laws criminalising deepfake fraud and non-consensual deepfake content. The United Kingdom’s Online Safety Act includes provisions addressing the harmful use of synthetic media. Despite these legislative developments, the global and borderless nature of the AI deepfake scam ecosystem makes enforcement extremely challenging — placing a significant burden on individuals and organisations to protect themselves through their own verification habits.

Establish a Family Emergency Code Word

Agree on a secret code word with your immediate family members that would only be known to you. If you ever receive a distressing call from an apparent family member in an emergency, ask for the code word before taking any action. An AI deepfake scam operator cannot provide a code word they don’t know. This single measure provides robust protection against the virtual kidnapping variant of the AI deepfake scam.

Always Verify Through a Separate Channel

If you receive any call, video call, or message requesting urgent money or sensitive information — regardless of how convincingly it appears to come from someone you know — hang up and call that person back on a number you already have saved. Never call back using a number provided during the suspicious contact. This independent verification step makes the AI deepfake scam essentially impossible to succeed against you.

Implement Verification Protocols in Your Business

For businesses, the AI deepfake scam risk requires formal procedural controls. Implement a requirement that any financial transfer above a threshold amount must be verified through a second, separately initiated communication channel — regardless of how convincing the original instruction appears. Train all finance staff to be aware of the AI deepfake scam threat and to treat any urgent payment instruction received through a video call or voice call as requiring independent verification before execution.

Limit Your Public Audio and Video Footprint

The less audio and video material of you that is publicly available online, the harder it is for criminals to build a convincing deepfake of you for use in an AI deepfake scam. Review your social media privacy settings and consider restricting video content to known contacts. This is particularly important for individuals who are prominent in their professional or public lives.

Be Sceptical of Celebrity Investment Content

Treat any video — however professionally produced — in which a celebrity, business leader, or public figure appears to endorse an investment platform with extreme scepticism. Before investing any money based on such content, verify through the person’s official verified accounts or official press releases that they have actually made the endorsement. The celebrity investment AI deepfake scam is almost always identifiable through this simple verification step. The same celebrity-endorsement playbook is dissected in our crypto investment scams guide.

Never Pay Through Untraceable Methods Under Pressure

No legitimate person — whether a family member, employer, financial institution, or government agency — will ever demand urgent payment through gift cards, cryptocurrency, or bank transfer without the opportunity to verify the request first. Any urgent payment demand through these channels, regardless of how convincing the requester appears, should be treated as a probable AI deepfake scam.

If you believe you have been targeted by or fallen victim to an AI deepfake scam, take the following steps as quickly as possible to limit the financial and personal damage.