How to spot AI fake news and deepfakes before you forward them

A "breaking news" card with a logo you almost recognise. A video of a leader saying something shocking. A voice note that sounds exactly like your cousin asking for money. AI tools have made fakes like these cheap and quick to make, and family WhatsApp groups are where many of them travel.

You cannot stop people from making fakes. You can stop yourself from passing them on. This guide gives you six checks, in the order we suggest using them, with the free tools to do each one. Steps 1, 2 and 5 take a few minutes on a phone and make a good first check. Steps 3 and 4 go deeper and can take longer, and the video tool in step 4 needs a computer.

Who this is for

Anyone who gets forwards: students, parents, office groups, society groups. You do not need to be technical. Most steps work on a phone. The video step works best on a computer.

Step 1: Pause when the message makes you angry or scared

Fake and misleading messages are built to trigger a strong reaction. Outrage, fear, urgency and "share before it gets deleted" are the usual hooks, because a person who reacts fast does not stop to check.

Try this habit: if your first reaction is "everyone must see this right now", wait ten minutes before doing anything. Pausing is a useful habit, not a verification - the checks below are what actually check the message.

Questions to ask yourself:

• Is it asking me to forward it to everyone?
• Does it say "forwarded as received" or "100% true" without naming a source?
• Does it push me to act (send money, click a link, share OTP details) quickly?

If you answered yes to any of these, treat the message as unconfirmed until you finish the steps below.

Step 2: Look at who is actually saying it

A forward is not a source. Look for a named news organisation, an official account, a date, and a place. A screenshot of a headline is easy to fake, so open the outlet's own website or app and look for the same story there.

WhatsApp marks messages that have been forwarded many times with a double-arrow label, to show they did not come from a close contact (WhatsApp blog). That label does not tell you whether a message is true or false. It tells you the message has passed through many hands, so check it before you add yours.

If the claim is about the Government of India

The Press Information Bureau (PIB) runs a Fact Check Unit. According to PIB's own Fact Check FAQ, you can reach it by:

• WhatsApp hotline: +91 8799711259
• Email: factcheck at pib dot gov dot in
• Website portal: factcheck.pib.gov.in

One limit to know: the unit only takes up claims related to the Government of India, its ministries, departments and public sector entities. A claim about a state government, a private company or a celebrity is outside its remit, so do not expect an answer from it on those.

Step 3: Reverse-search the picture

Many fake visuals are not new at all. They are old photos from a different event, put under a new caption. Others are AI images made to look like news photos.

Google Lens lets you search with an image to learn more about it (Google's help page). On a phone, you can open the picture, use Lens, and look at where else the same or a similar image appears.

What you are looking for:

• The same photo on older pages, with a different caption or location.
• A news report that names a different place or year.
• No other trace of the image at all - that alone does not prove the photo is fake or real, because a genuinely new photo may have no matches. Treat it as no confirmation and keep checking the source and context.

Step 4: For videos, check the frames

Videos are harder, because a clip can be real footage from another time or place, edited, or fully AI generated.

The InVID-WeVerify verification plugin (listed in the Chrome Web Store as "Fake news debunker by InVID, WeVerify and vera.ai") is a browser extension made for journalists and fact-checkers. Its project page on GitHub says it can split videos from several platforms into keyframes and run reverse image searches on Google, Baidu or Yandex. It is a desktop browser extension, so if a video matters, send the link to yourself and check it on a computer.

A simple routine:

• Get the video link, not a screen recording, if you can.
• Pull keyframes with the tool.
• Reverse-search two or three of them.
• Read the original upload date and the original caption.

Step 5: Search the claim on fact-checking sites

Before forwarding anything big, search the claim in plain words on Indian fact-checking sites such as Alt News, BOOM and Newschecker. They publish checks of viral claims, including AI-made content.

Two points to keep in mind:

• If you do not find a matching check, that is not proof the claim is true or false. Your search may have missed one, or nobody may have checked it.
• WhatsApp has a help page listing IFCN fact-checking organisations on WhatsApp. Look at it to see the tiplines you can message directly.

Step 6: Do not rely on spotting "deepfake tells"

Older advice said to look for odd hands, blurry edges, mismatched earrings or strange blinking. Do not rely on that. AI image and video tools keep improving, and a clip that looks perfect can still be fake, while a real clip can look rough.

Use context instead:

• Did any credible outlet report this event?
• Does the person's official account carry the statement?
• Is the audio from a real event, or is there only one clip and no other angles?

Be extra careful when money is involved. If a voice note or a call from someone you know asks for urgent money, do not act on it. Contact the person on a number you already have saved, or ask someone who is with them. You can also agree on a code word with your family that only you know.

A quick worked example (made up for illustration)

Imagine a message arrives: a news-style card says a new rule starts tomorrow and "everyone must share this today". Here is how the six steps play out:

• Pause: the urgency and "share today" are red flags.
• Source: no outlet is named. The rule seems to be about a central government scheme, so it is a candidate for PIB.
• Image: the card has no photo, so Lens does not help.
• Video: none attached.
• Search: you type the claim into a search engine plus "fact check". No matching check turns up, which proves nothing either way.
• Context: you open the official ministry website and see no such notice.

Result: you do not forward it. If you are curious, send the message to the PIB WhatsApp hotline and wait for their reply.

What to do when you find a fake

• Do not forward it, even to say "this is fake". A screenshot with a "FAKE" note can still spread without the note.
• Reply in the group with a short message and a link to the fact-check or the official page.
• Use WhatsApp's report option for the message or the sender if it is clearly harmful.
• If someone in your group keeps forwarding such messages, send them the checklist privately and politely. Anyone can be taken in by a convincing fake.

The one-line version

Strong emotion, no source, and "forward to everyone" means do not forward. Check the source, run a reverse image search, search the claim on a fact-checking site, and look at the official account. A quick first check takes a few minutes. Deeper checks take longer, and you can leave a message unforwarded while you do them.

Frequently asked questions

Can AI detectors tell me if a picture or video is fake?

Treat any single detector result as one clue, not a verdict. Combine it with the source, the date and the context.

Is a forwarded message with the double-arrow label always fake?

No. The label only shows the message has been forwarded many times. Check it like any other message.

Can I send any claim to the PIB hotline?

Only claims related to the Government of India and its ministries, departments and public sector entities. Others are outside its remit.

What if I cannot find a fact-check?

Then you have no confirmation. Do not forward until a credible outlet or an official source confirms it.

We share practical AI explainers like this, written for India, in our free community. Join at nudgedaily.in.

Sources

All of these pages were fetched or searched on October 24, 2026.

• PIB Fact Check Unit FAQ
• WhatsApp Help Center: IFCN fact-checking organizations on WhatsApp
• WhatsApp blog: Keeping WhatsApp Personal and Private
• Google Search Help: Search with an image on Google
• InVID-WeVerify plugin on the Chrome Web Store
• InVID-WeVerify plugin on GitHub (AFP-Medialab)