Spot AI fakes: pictures, videos, voices and scams
AI can fake a photo, a video or a loved one’s voice, and the tell-tale glitches are disappearing. Here’s how to check what’s real, starting with where it came from.
In 30 seconds
- Many fakes are now too good to spot by eye, so check where they came from instead.
- Feeling angry, scared or rushed? Pause. Fakes and scams are built to make you react before you think.
- A call from family asking for money? Hang up and call back on a number you know.
A photo of a flooded high street is going round the family group chat. A video shows a famous presenter backing an investment. Your phone rings, and it’s your daughter, in tears, needing money now. Any of these could be made with AI. You can’t always tell by looking or listening, but you can almost always check.
Why hunting for glitches isn’t enough
You may have heard the old tips: look for extra fingers, garbled writing or odd shadows. They used to work. But AI tools keep improving, and many of those glitches have gone. A Jargon busterDeepfake: A fake video, image or voice recording made with AI to look or sound like a real person. can look and sound convincing, while a real photo can look odd for innocent reasons.
So treat a glitch as a bonus clue, not a test. Spotting one is useful, but not spotting one proves nothing. Detector tools and chatbots can call a real picture fake, or a fake one real. And while some apps label AI-made content, a missing label doesn’t mean something is real.
| Instead of | Try |
|---|---|
| Zooming in to count fingers | Finding out who posted it first, and where |
| Trusting it because it looks real | Searching with the picture to find older copies |
| Asking a chatbot whether it’s fake | Checking trusted news sites or a fact-checker |
| Sharing it while everyone’s talking about it | Waiting until you’ve checked |
Check the source, in four steps
Before you believe it or share it
- PauseNotice how it makes you feel. Fakes are built to make you angry, scared or rushed, so you react before you think.
- Check who posted itIs it a news organisation or an official account, or a stranger, a new account or a forwarded message? Check the account is genuine, and find the original post.
- Search with the pictureA Jargon busterReverse image search: A search that starts from a picture instead of words. It finds where else the picture appears online, which can show where it first came from. finds other copies of a picture, or of a screenshot from a video. If it appeared years ago, or somewhere else entirely, it’s been reused.
- See who else reports itBig news gets covered. If no trusted news site has it, be wary. Fact-checkers such as Full Fact and BBC Verify investigate viral claims.
Try it yourself 5 minutes
- Pick a striking photo from the news or your social media feed this week, and save it or copy its link.
- Open a search engine that can search with a picture, often through a camera icon in the search box, and add the photo.
- Look through the results for earlier copies, and for where it first appeared.
- Ask yourself: is the photo older than the story it’s attached to? Do trusted news sites use it the same way?
Cloned voices and scam messages
Jargon busterVoice cloning: Using AI to copy someone’s voice from a short recording, so it can say things they never said. can copy a voice from a short clip, such as a video posted online. Scammers can use it to sound like a relative in trouble, or a boss who needs a payment made today. Jargon busterPhishing: A scam message pretending to be from someone you trust, to trick you into sharing details or paying. messages have changed too: AI can write them with perfect spelling, so judge the request, not the writing.
- The call“Mum, it’s me. I’m in trouble and I need money now.”
- Ask for the code wordOne only your family knows
- Hang upEven if it sounds just like them
- Call backOn a number you already have
Write a short, friendly message to my family suggesting we agree a code word to use if one of us ever calls or messages asking for money in an emergency. Explain in one sentence that AI can now copy voices. Keep it calm and practical, not scary. Suggest we choose the word together in person, and never write it in a group chat or post it online.
If you’ve been caught out
Don’t be embarrassed: these scams are built to fool careful people. If you’ve shared bank details or sent money, call your bank straight away, using the number on the back of your card.
Then report it. In England, Wales and Northern Ireland, use Report Fraud, which replaced Action Fraud. In Scotland, call Police Scotland on 101. Forward scam texts to 7726, which is free, and report fake posts to the site or app where you saw them.
For the privacy settings worth checking, read is AI safe to use? And because chatbots can be confidently wrong too, see why AI makes things up.
Check yourself
3 quick questions nothing is savedSpotted a mistake? Tell us and an editor will check it.
