How to spot AI-generated videos: key methods to know

NEW YORK: Social media is flooded with videos created with artificial intelligence (AI), but one important clue to identifying fake videos is whether the quality of the video is so poor that it seems like it was made with an old mobile phone or a weak camera.

According to experts, AI video-making tools have become so advanced in the last 6 months that it is now becoming increasingly difficult for us to distinguish between the real and the fake.

Experts say that the AI ​​videos that are currently deceiving us are usually of low quality.

The reason is that when a video is intentionally blurred or kept in low resolution, small flaws such as strange movements, unnatural facial expressions, or background errors are hidden.

For example, a fake video of rabbits jumping on a trampoline was viewed more than 240 million times on TikTok.

A video of two people ‘falling in love’ on the New York subway also fooled millions of people.

And a video of a fake priest, in which he was giving a speech against the wealthy class, was viewed by millions of people as real.

One thing all these videos had in common was that they were all low-quality footage.

However, another hallmark of fake videos is their length. He says that most videos are only 6 to 10 seconds long because making longer videos is expensive and difficult.

Similarly, low resolution and high compression can also be a sign, as fraudsters often deliberately blur videos to hide flaws.

Experts warn that this advice will not always work because AI is improving rapidly.

Drexel University professor Matthew Stam says that within 2 years these visual signs will disappear and we will not be able to trust our eyes, but there is still hope.

Digital forensics experts are working on digital fingerprints that can be used to determine whether a video is real or fake. In the future, it is possible that cameras themselves will add verification data (metadata) to the video to prove its authenticity.

Real Solution?

According to digital literacy expert Mike Caulfield, the solution to the problem is not technology, but our attitude.

He said that we have to learn to view videos like writing, not just believe what we see, but examine its source, context, and the person posting it.

Mike Caulfield said that it has now become dangerous to assume that a video is real just by looking at it, and to know the truth, it is important that we ask where this video came from?

He said that we should see who uploaded this video and has any reliable source confirmed it?

As Professor Stam says, this is the biggest information security challenge of the 21st century, but we are not ready to give up yet.