What is a deepfake scam?
A deepfake scam uses AI-generated audio or video to impersonate a real person — a family member, celebrity, CEO, or public official — to deceive victims into transferring money or sharing sensitive information.
Last reviewed: 10 June 2026
Explanation
Deepfake technology uses machine learning to synthesise realistic audio and video of a person saying or doing things they never did. What once required studios and skilled effects artists can now be produced by free or cheap consumer tools from a few minutes of source material — a social media video, a public speech, or a phone conversation.
In financial fraud, deepfakes are used in several ways. 'Virtual kidnap' calls use a cloned child's or grandchild's voice to demand ransom. CEO fraud deploys a CFO's convincing voice in an urgent call requesting a wire transfer. Celebrity deepfake videos promote fake investment platforms and crypto giveaway scams to millions of viewers simultaneously.
At the personal level, deepfakes compound romance and sweetheart scams: a fake video call with an AI-generated face convinces a victim they have seen their online partner, dispelling doubts. At the corporate level, deepfake video calls have been used in boardroom-level fraud where executives are impersonated to authorise transactions.
Defences include establishing out-of-band verification — a family code word, a known phone number to call back on, or a challenge question agreed in advance. For businesses, multi-person authorisation for high-value transactions and a strict policy of never changing payment details based on a video or call alone are essential.
Common red flags
- A video or audio call from a known person making an urgent, unusual financial request
- Celebrity or public figure videos promoting investment opportunities with guaranteed returns
- A 'live' video in which the face is stiff, lighting inconsistent, or lip-sync slightly off
- Emotional pressure to act immediately without telling anyone else
- The person in the video is unable to respond naturally to unexpected questions
- Payment or credential requests that bypass normal verification procedures
What to do now
- Hang up and call the person back independently on a known, saved number
- Use an agreed family or corporate code word to verify genuine emergencies
- Do not transfer money or share credentials based on a video or voice call alone without secondary verification
- Report AI-generated impersonation to the platform and to your national fraud authority
- For businesses, review payment authorisation procedures to require multi-person sign-off
Frequently asked questions
How good are deepfakes now — are they detectable?
Quality varies. Some deepfakes are clearly artificial; others are convincing enough to pass a brief real-time call. Detection tools exist but are not perfect. Behavioural verification — asking unexpected questions, using code words — is more reliable than trying to spot visual artefacts.
Is it illegal to make a deepfake of someone?
Laws vary by jurisdiction. Many countries now have or are introducing laws against non-consensual intimate deepfakes and fraudulent deepfakes used in crime. Using a deepfake to commit fraud is already criminal under existing fraud statutes in most places.
Sources
- FTC — Scammers Use AI to Enhance Family Emergency Schemes — Federal Trade Commission
- FTC — Fighting Back Against Harmful Voice Cloning — Federal Trade Commission
- FBI IC3 — Criminals Use Generative AI to Facilitate Financial Fraud — FBI Internet Crime Complaint Center
- AARP — Fraud Watch Network — AARP