Definition
A video shows a well-known leader announcing something shocking. The voice is right and the face is right. The speech never happened. How is that possible?
Misinformation is false or misleading information shared by people who believe it. Disinformation is false information created and spread on purpose to deceive. A deepfake is realistic fake audio, image or video of a real person, made with AI.
In simple words
A rumour in a village once took days to travel from house to house, and changed a little with each telling. Today the same rumour reaches a million phones in an hour, with a photograph attached. And the photograph itself can now be manufactured. The rumour is old. Its speed, and the forged evidence that travels with it, are new.
Why it matters
False information has led to panic, violence, lost savings and damaged reputations. Everyone who forwards a message is part of how it spreads, so everyone has a share in stopping it. Understanding how falsehoods work is the first defence.
Key terms
- Misinformation
- False information shared without intending harm.
- Disinformation
- False information made and spread deliberately to mislead.
- Deepfake
- AI-made audio, image or video that shows a real person saying or doing something they never did.
- Cheapfake
- Real content that has been simply edited or given a false caption.
- Confirmation bias
- The tendency to believe what agrees with what we already think.
- Virality
- Spreading very quickly from person to person.
Explained in detail
Falsehood travels faster than truth
In 2018 researchers at the Massachusetts Institute of Technology studied about 126,000 news stories shared on a large social network over eleven years. False stories spread further and faster than true ones in every subject. A false story reached 1,500 people about six times more quickly than a true one. The main cause was not automated accounts. It was people, who were more likely to share the false stories because they were more surprising and stirred stronger feelings.
Why sensible people believe false things
- Confirmation bias: we accept without checking whatever fits what we already believe.
- Repetition: a claim heard many times begins to feel true, even to people who first doubted it.
- Trust in the sender: a message from an aunt or a close friend arrives with their credibility attached.
- Strong feeling: anger and fear make us act before we think.
- A grain of truth: the most effective falsehoods are built around something real.
How a deepfake is made
A deepfake uses the same generative AI that creates harmless pictures. For a face swap, a model studies many images of a person's face and learns to paint that face, with matching expressions, on to another person's body in each frame of a video. For a voice clone, a model listens to a sample of someone's speech and then reads any text aloud in that voice. A third method changes only the mouth, so that a real video of a person appears to speak new words. A few years ago this needed thousands of photos and expert skill. Today a phone app, one photograph and a few seconds of sound can be enough.
What deepfakes are used for
| Use | What happens | Who is harmed |
|---|---|---|
| Fraud | A cloned voice or face asks for money | Families and companies |
| Harassment | A person's face is placed in embarrassing or indecent content without consent | Mostly women and girls |
| Politics | A leader appears to say something they never said | Voters and public trust |
| Panic | A fake image of a disaster or attack goes viral | Whole communities |
Studies of deepfake videos found online have repeatedly shown that the great majority target women, placing their faces in content they never agreed to. In elections around the world, including India's in 2024, AI-generated videos and cloned voices of politicians were circulated, among them the voices of leaders who were no longer alive.
The commoner trick needs no AI
Most false content is far simpler than a deepfake. A real photograph of a flood from another country and another year is posted as "our city this morning". A real video is cut so that the sentence explaining a remark is removed. A clip is slowed down to make a speaker seem unwell. These cheapfakes are quick to make and hard to disprove with technology, because nothing in the picture is fake. Only the caption or the context is false.
What you can do
- Pause before sharing anything that makes you angry, afraid or triumphant.
- Check where it came from and whether reliable outlets report it.
- Do not forward "just in case it is true". Forwarding is how it spreads.
- Never create or share a fake image or voice of a real person. It can do lasting harm and may be a crime.
- If you or someone you know is targeted, save the evidence, report it on the platform and at cybercrime.gov.in, and tell a trusted adult.
How it works
| Type | Who made it | Intent | Example |
|---|---|---|---|
| Misinformation | An ordinary person | Believes it is true | Forwarding a false health tip to help the family |
| Disinformation | Someone with a motive | To deceive | A made-up story to damage a rival |
| Deepfake | Someone using AI tools | Usually to deceive or harm | A cloned voice asking for money |
| Cheapfake | Anyone with a phone | To mislead | An old photo with a new, false caption |
Worked example
One evening the phone rings from an unknown number. The voice sounds exactly like your cousin Rohan: 'I've had an accident. Please send ₹20,000 to this UPI ID right now. Don't tell anyone, and don't call back, my phone is about to die.' AI tools can now copy a voice from a short recording, such as a video posted online, so a familiar voice is no longer proof of who is speaking.
Look at the warning signs together: urgency, secrecy, a new payment ID and a reason why you cannot call back. The decision is simple: do not pay. Hang up and call Rohan on the number you already have saved, not the number that called, or call a family member who is with him. In this case Rohan answers from college; he is fine, and the call used a cloned voice. Report it on the National Cyber Crime Reporting Portal, cybercrime.gov.in, or by calling the helpline 1930.
Try it yourself
How a false video spreads through a town
An old video of a crowd is posted with a new caption claiming trouble in the local market.
What kind of falsehood is this?
Think about who made it and why.
A group invents a story about a rival candidate and pays for it to be spread
A photo of last year's flood is posted as today's
A neighbour forwards a home remedy that does not work, believing it will help
A video uses AI to make an actor appear to promote a betting app
A student shares a wrong exam date heard from a friend
A fake website publishes false cures in order to sell a product
A familiar voice asks for money: safe step or risky step?
The caller sounds exactly like a cousin and says it is urgent.
Share an OTP because the caller says it will help your cousin
Hang up and call back on the number you already have saved
Ask a question only the real person would know
Report the call on cybercrime.gov.in or the helpline 1930
Call back on the number that just called you
Send the money quickly because it sounds urgent
A new case: the 'bank officer' on a video call
Decide your steps before anything like this happens.
- A video call shows a person in a uniform who says they are from your parent's bank. They ask for card details 'to stop a fraud'. Write what you do, step by step.
- Write the number you would use to check with the real bank, and where you would find it.
- Agree with your family on one question that only family members could answer.
Think about it
- Why is a face or a voice on a screen no longer enough to prove who someone is?
Common mistakes
- Thinking only strangers spread false information. Most of it arrives from people we trust.
- Treating a video as proof because "you can see it happening".
- Making a fake image of a friend or teacher as a joke.
Check yourself
Summary
- Misinformation is shared by mistake; disinformation is spread on purpose.
- Deepfakes forge a real person's face or voice; cheapfakes mislead with real content and a false context.
- Pause, check and refuse to forward what you cannot confirm.
Finished reading?
Sources: Civic Online Reasoning (Digital Inquiry Group) · National Cyber Crime Reporting Portal (India)