AI-generated videos have become surprisingly convincing. A few years ago, it was relatively easy to tell when a video had been digitally manipulated. Today, AI tools can create realistic faces, voices and even entire conversations that can look and sound like the real thing.
Recently, I came across a video that appeared to show Bollywood actor Salman Khan speaking about the reported demise of Nims Dai. At first glance, the video looked convincing. The face looked familiar, the voice sounded similar and the message appeared believable.
But this is exactly where we need to be careful.
A video showing a famous person saying something shocking does not automatically mean that the person actually said it. AI-generated videos, commonly called deepfakes, can put words into someone’s mouth that they never actually spoke.
So how can you tell whether a video is real or AI-generated?
1. Look closely at the person’s mouth
One of the easiest warning signs is the way the person’s lips move.
In AI-generated videos, the mouth may not perfectly match the words being spoken. You might notice strange lip movements, unnatural pauses or moments where the mouth appears slightly out of sync with the audio.
Watch the video several times, especially around difficult words or fast sentences.
If the audio says one thing but the person’s lips appear to form something different, be suspicious.
2. Pay attention to the eyes
Eyes can reveal a lot.
AI-generated faces sometimes have unnatural blinking patterns. A person may blink too frequently, barely blink at all or have strange-looking eye movements.
Look at whether the eyes move naturally when the person is speaking. Also check whether the direction of the person’s gaze suddenly changes without a logical reason.
These details can be difficult for AI systems to reproduce perfectly.
3. Check the face for strange movements
A person’s face naturally changes while they speak.
Their forehead moves, cheeks change shape, eyebrows react and small muscles around the eyes move constantly.
In a fake AI video, these movements can sometimes look strangely smooth or disconnected.
For example, the mouth may move normally while the rest of the face remains unusually still. Or the facial expression may change suddenly without matching what the person is saying.
If the face looks almost perfect but somehow feels “off,” take a closer look.
4. Listen carefully to the voice
AI voice cloning has become extremely advanced, but it is not always perfect.
Listen for unusual pronunciation, strange pauses, robotic rhythm or words that sound slightly distorted.
Sometimes an AI-generated voice sounds like the person but lacks the natural emotion you would expect.
A famous actor delivering a serious message may normally have changes in tone, emphasis and breathing. An AI-generated voice can sometimes sound unusually consistent from beginning to end.
5. Look for unnatural skin and hair
Zoom into the person’s face.
AI-generated videos can sometimes produce strange skin textures, especially around the cheeks, ears, hairline and neck.
Hair may appear to merge into the background. Skin may look excessively smooth. Small details may disappear when the person moves.
Pay particular attention to the ears, teeth and hair because these areas can expose imperfections that are difficult to notice when watching the video normally.
6. Check the hands and body
Don’t only look at the face.
AI video generators can struggle with hands, fingers and body movements.
If the person raises their hands, look at the fingers. Do they have the correct number? Do they move naturally? Do their hands suddenly change shape?
Also check whether the person’s head, neck and shoulders move naturally together.
A video can have a convincing face while the rest of the body gives away the manipulation.
7. Look at the background
The background can provide another important clue.
Check whether objects behind the person remain consistent throughout the video.
Look for things such as:
- Objects changing shape
- Text becoming distorted
- Furniture moving slightly
- Unusual shadows
- Blurry edges around the person
- Objects appearing or disappearing
- Background details that don’t make physical sense
AI-generated videos sometimes focus heavily on making the person look realistic while the surrounding environment contains small mistakes.
8. Don’t trust a video just because it looks professional
This is one of the biggest mistakes people make.
A video can have excellent lighting, subtitles, background music and professional editing and still be completely fake.
In fact, scammers often make fake videos look professional because it makes them appear more trustworthy.
A video claiming that a celebrity has died, endorsed a product, promoted an investment or made a shocking statement should immediately make you look for confirmation from reliable sources.
The quality of the editing is not proof that the video is genuine.
9. Check where the video originally came from
Before believing a viral video, find out who posted it first.
Was it uploaded by the celebrity’s verified account? Was it published by a reputable news organization? Or did it come from an unknown Facebook page, TikTok account, YouTube channel or random social media profile?
This matters enormously.
If a video claims that Salman Khan made an important statement, for example, look for the same statement on his official social media accounts or in reputable news reports.
If nobody credible is reporting it, that is a major reason to be skeptical.
10. Search for the story outside social media
This is probably the most important step.
If a video claims that a famous person has died or that a celebrity made a major announcement, don’t rely on the video itself.
Search for the claim.
Look at multiple reputable news organizations and the person’s official social media accounts. If the event is genuinely significant, there will usually be independent reporting or some form of official confirmation.
For example, if you see a video appearing to show Salman Khan talking about Nims Dai’s reported death, don’t immediately share it because the video looks convincing.
First ask:
“Is this actually being reported by reliable sources?”
That simple question can prevent misinformation from spreading.
If this article helped you, please consider supporting our work. Every small contribution keeps Abijita.com independent and running.
Why AI Fake Videos Are Becoming So Dangerous
The biggest problem is that AI-generated videos are no longer limited to obvious computer-generated faces.
Modern AI tools can imitate someone’s appearance, voice and mannerisms with remarkable accuracy. Someone can potentially create a video of a politician making a statement, a celebrity promoting a product or a public figure discussing an event that never happened.
This creates a serious misinformation problem.
The more emotional the video is, the more likely people are to share it without checking.
Death announcements, breaking news, political statements, celebrity scandals and financial opportunities are particularly powerful because they trigger an immediate emotional reaction.
That’s exactly why you should slow down when a video makes you angry, scared or extremely excited.
