How do I protect myself from AI-powered scams like voice cloning and deepfakes?
Establish a verbal code word with close family for emergency requests, and treat any call creating urgency around money with scepticism regardless of how authentic the voice sounds.
Last reviewed: 10 June 2026
Explanation
Artificial intelligence has lowered the barrier to creating convincing fake audio and video of real people. Voice cloning requires only a short sample of someone's voice — obtainable from a voicemail, a social media video, or a public speech — to generate audio that sounds convincingly like that person saying anything. Deepfake video has become accessible enough to be deployed in scams. These technologies are being used to supercharge existing scam formats: grandparent scams with cloned grandchildren's voices, fake emergency calls from 'kidnapped' family members, and corporate fraud using a CEO's voice or face.
The most effective defence is a verification protocol that bypasses the audio or video entirely. Agree a secret code word with close family members — a word that a scammer would not know — that can be used to confirm genuine identity in any emergency request involving money. If the caller cannot produce the code word, hang up and call the real person on their known number using a second device.
For business contexts, implement the same verbal-confirmation rule that protects against business email compromise: any request from an executive to transfer funds must be confirmed by calling that executive on their known phone number, regardless of how convincing the original request sounded. A scammer impersonating a CFO on a call to authorise a wire transfer is defeated the moment an independent confirmation call is required.
Be especially sceptical of unsolicited contact from investment advisers, celebrities, or public figures promoting opportunities via video — deepfake celebrity investment scams are common and are often promoted through social media. Verify any investment opportunity through the regulator register, not through the credibility of the face promoting it.
Common red flags
- Unexpected call from a family member saying they are in danger and need money immediately
- Caller sounds like someone you know but the request is unusual and urgent
- Video call or social media video of a celebrity promoting a specific investment platform
- Executive calling to authorise an urgent wire transfer outside normal channels
- Slight audio artifacts, robotic pacing, or lip-sync inconsistencies in video
- Call or message says to tell no one else about the emergency
What to do now
- Establish a secret code word with close family for genuine emergencies
- For any urgent money request, hang up and call the real person back on their known number
- Apply verbal confirmation rules to all executive payment authorisations in your workplace
- Treat all celebrity-endorsed investment offers with extreme scepticism regardless of the medium
- Report AI-enabled scam attempts to the FTC at ReportFraud.ftc.gov
- Visit /family-safety to share these protocols with family members
Frequently asked questions
How realistic is voice cloning today?
Consumer-accessible voice-cloning tools can produce convincing results from short audio samples. The output is not always perfect — there may be unnatural pacing or audio artifacts — but in a stressful situation a person may not notice. A code word removes the need to distinguish real from fake by ear.
Can deepfake video be used in real-time calls?
Real-time deepfake video is technically possible and has been used in documented fraud cases. Video alone should not be treated as proof of identity for high-stakes decisions. A code word or calling back on a known number is a more reliable verification method.