Skip to content

Deepfake and AI Voice Scams: How to Stay a Step Ahead

A panicked call that sounds exactly like your brother. A video call from a 'match' whose face looks almost right. Here's how deepfake and AI voice scams work.

By The Kindred Team

6 min read

Tower Bridge over the River Thames in London, illustrating a guide to deepfake and AI voice scams

Your phone rings. It's your younger brother, and he sounds terrified. He's been in an accident, or arrested, or stranded somewhere, and he needs money right now, please, don't tell Mum. The voice is his. The way he says your name is his. Your heart is pounding before your brain has even caught up.

Except it isn't him. It's software, trained on a few seconds of his voice from a video he posted online.

Deepfake and AI voice scams sound like science fiction, but the tools behind them are now cheap and widely available. That doesn't mean you need to panic or stop posting anything ever again. It means updating a few habits so that a convincing voice or face is never enough, on its own, to move your money or your trust.

What deepfakes actually are

A deepfake is audio, video or an image that has been generated or altered using AI to make it look or sound like a real person said or did something they didn't. Wikipedia's deepfake article has a thorough overview of the technology and its history.

For scams, three types matter most:

  • Voice clones. AI copies someone's voice from a short sample and uses it in calls or voice notes.
  • Face swaps and fake video. A scammer's face is replaced with someone else's on a video call or in a recorded clip.
  • AI-generated faces and photos. Completely fake people, used for profile pictures on social, dating and professional sites.

Each one is designed to borrow trust. Your trust in a family member's voice. Your trust in seeing someone's face. Your trust in a friendly profile photo.

Seeing used to be believing. Now, checking is believing.

How these scams usually play out

The family emergency call

The most common voice-clone scam is a version of an old trick: a relative in trouble who needs money urgently. The AI voice just makes it far more convincing. These calls often come with background noise, crying or a second person claiming to be a police officer or lawyer, plus strict instructions not to tell anyone.

The boss or colleague request

At work, scammers have used cloned voices and faked video calls to impersonate senior staff and push employees into urgent payments or sharing sensitive information.

The "real" person behind a fake profile

On social and dating apps, a scammer might use AI-generated photos, or even a short face-swapped video call, to "prove" they're real. This is a newer twist on catfishing, which we cover in how to spot a catfish.

Celebrity endorsements

Fake videos of famous people promoting investment schemes, giveaways or products are common on social media. If a celebrity appears to be promising you riches, assume it's fake.

Fake intimate images

AI can also be used to create fake explicit images of real people, sometimes used for harassment or extortion. If this happens to you, please know that it's not your fault and help exists. Our guide on sextortion and what to do covers next steps and support.

Signs a voice or video might be fake

Detection is getting harder as the tools improve, so treat these as clues, not proof.

In audio and voice notes

  • Flat or oddly even emotion, even when the words are dramatic.
  • Strange pauses, clipped words or robotic edges to some sounds.
  • Background noise that loops or sounds pasted on.
  • They avoid answering unexpected questions, or the conversation feels scripted.

In video calls

  • Edges that flicker around the face, hair, glasses or jawline, especially when they move.
  • Lip movements slightly out of sync with the words.
  • Odd lighting on the face that doesn't match the room.
  • Reluctance to move. Ask them to turn their head, wave a hand in front of their face or hold up a specific number of fingers. Face-swap tools often struggle with this.

In photos

  • Mismatched earrings, warped backgrounds, strange hands or teeth, glasses that blend into skin.
  • No trace of the person anywhere else online, or the same face appearing under different names. Our reverse image search how-to shows you how to check.

The habits that beat deepfakes

Because detection is unreliable, the best defence isn't spotting the fake. It's having habits that make the fake irrelevant.

1. Hang up and call back

If someone calls asking for money or urgent action, hang up and call them back on the number you already have saved. If it's a colleague, call through the official company directory. A real person will understand. A scammer will lose their grip the moment you step outside their call.

2. Agree on a family code word

Pick a word or question only your real family knows, something not findable on social media. If anyone calls claiming to be family and asking for money, ask for the code word. We explain how to set this up in our guide on protecting parents from online scams.

3. Never let urgency decide

Scammers rush you because thinking is their enemy. "Don't hang up", "don't tell anyone", "it has to be in the next hour" are red flags in themselves.

4. Verify through a second channel

Got a voice note from a friend asking for money? Message them separately, call them, or ask a mutual friend. One channel can be faked. Two independent channels are much harder.

5. Be thoughtful about what you post

You don't need to disappear from the internet. But you might choose to keep long videos of yourself talking on a private account, or be more selective about public voice notes. The less clean audio of you that's public, the harder it is to clone.

6. Treat celebrity money offers as fake by default

Real celebrities don't personally DM you about crypto, and they don't do giveaways that require a "small processing fee".

The UK's National Cyber Security Centre's guide to phishing scams is a useful reminder that many of these scams are classic social engineering with a new coat of paint.

Staying safe on social and friendship apps

When you're meeting new people online, combine a video call with patience. A video call is still a valuable step, just not the only one. Meet in public, let trust grow over time, and keep money completely out of new relationships.

On Kindred, links, phone numbers and emails are blocked for the first 48 hours of a new match, so it's harder for anyone to rush you to another app, and the scam guard in chats warns you about risky messages. Report and block are always reachable from any chat, never paywalled. You can read more on our safety page.

If you think you've been targeted

  1. Stop and verify through a separate channel before doing anything else.
  2. Don't send money or codes.
  3. If you've already paid, call your bank immediately.
  4. Save the evidence: call logs, voice notes, videos, usernames.
  5. Report it to the platform and the police, and follow the steps in what to do if you've been scammed.
  6. Warn your circle. If someone cloned your voice or face, let friends and family know so they don't fall for it.

Communities are surprisingly good at this. A group chat where people post "is this real?" before reacting can catch fakes faster than any software. Our community hub has more on building groups where people look out for each other.

Your challenge: tonight, message your closest family group and suggest a code word. Make it funny. Make it memorable. Then agree that nobody ever, ever sends money after a panicked call without using it first.