A deepfake is a fake video, photo or audio clip created using artificial intelligence to make someone appear to say or do something they never did. The term combines “deep learning” and “fake”.

How are deepfakes made?

AI models are trained on photos, videos or voice recordings of a person. They learn the person’s face and voice patterns and can then generate new, realistic-looking content. Tools to do this have become easier to use, which means deepfakes are becoming more common.

How deepfakes are misused

• Fraud: scammers clone a family member’s voice to make an urgent call asking for money.
• Fake endorsements: videos of celebrities appearing to promote investment schemes or products.
• Misinformation: fake clips of politicians or public figures.
• Harassment: fake images used to embarrass or blackmail people.

How to spot a deepfake

• Faces: unnatural blinking, blurry edges around the face, odd skin texture or lighting.
• Lips: mouth movements that do not quite match the words.
• Hands and details: strange fingers, jewellery or teeth.
• Voice: flat emotion, odd pauses or robotic tone.
• Context: is the message urgent, unusual, or asking for money or secrecy?

Note that deepfakes are improving quickly, so these signs are not always visible.

Protect yourself and your family

• Verify through another channel: if a relative calls asking for money urgently, hang up and call them back on their known number.
• Agree on a family code word for emergencies.
• Do not trust investment offers just because a famous person appears to endorse them.
• Limit how many clear photos and voice clips you share publicly.
• Report fraud on the cybercrime helpline 1930 or cybercrime.gov.in.

The fight back

Researchers and technology companies are developing detection tools and content labelling so that AI-generated media can be identified, and governments are framing rules to tackle harmful deepfakes.