Deepfakes are AI-generated or AI-altered images, videos or audio that make someone appear to say or do something they did not. Spotting them is getting harder, so rely on context checks: the source, whether trusted outlets report it, reverse image search, and unnatural details. For urgent calls asking for money, always verify through another channel.
What is a deepfake?
A deepfake is synthetic media created with AI to imitate a real person or event. It might be a video of a politician saying something they never said, a cloned voice of a family member asking for money, or a fake image of a disaster. The technology also has legitimate uses in film and accessibility, but the harms are serious: fraud, harassment, reputational damage and misinformation.
Why visual clues are no longer enough
Early deepfakes had obvious flaws. Modern tools produce very convincing results. Visual and audio clues still help, but context checks are now more reliable than spotting glitches.
Possible signs in images
- Inconsistent lighting or shadows.
- Odd hands, ears, teeth or jewellery.
- Garbled or nonsensical text on signs and clothing.
- Backgrounds that blur or warp strangely.
- Skin that looks too smooth or plastic.
Possible signs in video
- Lip movements that do not quite match speech.
- Unnatural blinking or facial expressions.
- Edges of the face flickering, especially when the head turns.
- Lighting on the face that does not match the scene.
Possible signs in audio
- Flat emotion or unusual pacing.
- Odd breathing patterns or none at all.
- Background noise that cuts in and out.
- Pressure for urgent action, combined with a familiar voice.
Context checks that work
- Check the source. Who posted it first? Is it a trusted outlet or an anonymous account?
- Look for confirmation. Would a real event this important be reported by multiple reliable news organisations?
- Reverse image search. Search the image to find where it first appeared and whether it has been altered.
- Look for the full context. Short clips can be misleading even when real.
- Pause before sharing. Emotional, shocking content is designed to spread quickly.
Protecting your family from voice scams
Criminals can clone voices from short clips posted online and call relatives claiming an emergency. Protect yourself:
- Agree a family code word for urgent requests.
- Hang up and call back on a number you know.
- Be suspicious of requests for money transfers, gift cards or secrecy.
- Limit public videos and voice notes of children.
If you are targeted
If someone creates a deepfake of you or your child, save evidence, report it to the platform, and contact the relevant cybercrime authority. In Pakistan, complaints can be made to the national cybercrime authorities. Talk to someone you trust; being targeted can be distressing, and support is available through our wellbeing team.
Frequently asked questions
Can AI detection tools reliably spot deepfakes?
Detection tools help but are not fully reliable, as generation methods change quickly. Combine them with context checks.
Is making deepfakes illegal?
Laws vary by country. Using deepfakes for fraud, harassment, defamation or non-consensual intimate imagery is illegal in many places.
How can I teach children about deepfakes?
Look at examples together, discuss why people make them, and practise the context checks above.
