What Are AI Deepfakes And How To Detect Them - Here's All You Need To Know

Computers are becoming increasingly capable of simulating reality. Artificial intelligence (AI)-generated media has been generating headlines, particularly films tailored to impersonate someone, making it appear as if they're saying or doing something they're not.

A Twitch streamer was discovered on a website notorious for creating artificial intelligence-generated pornography of his peers. A group of teenagers in New York filmed their principal making racist slurs and threatening pupils.

Instagram

In Venezuela, created videos are allegedly used to spread political propaganda. In all three situations, the AI-generated films were created with the intention of convincing people that someone did something, which in reality they did not. Deepfake is a term for this type of content.

What is Deepfake?

Deepfakes employ artificial intelligence to create entirely new video or audio with the objective of displaying something that did not occur in reality.

The term "deepfake" refers to the underlying technology ¡ª deep learning algorithms ¡ª which can be used to create fake material of real people by teaching themselves to solve issues using enormous amounts of data.

How is Deepfake different from other online manipulations?

Deepfakes aren't just any old bogus or deceptive image. The AI-generated pope in a puffy jacket and the phony scenes of Donald Trump being detained that spread shortly before his indictment are both AI-generated, but not deepfakes. (When images like these are paired with false information, they are referred to as "shallowfakes.")

The element of human input distinguishes a deepfake. When it comes to deepfakes, the user only gets to decide if what was created is what they want or not at the very end of the generation process; aside from tailoring training data and saying "yes" or "no" to what the computer generates after the fact, they have no say in how the computer chooses to make it.

How to detect a deepfake?

AP

Deepfakes can be identified by a few indicators: Do details appear hazy or obscure? Look for skin or hair issues, as well as faces that appear to be blurrier than the environment in which they are placed.

The focus may appear abnormally soft. Is the illumination artificial? Deepfake algorithms frequently keep the lighting from the clips used as models for the fake video, which is a poor match for the lighting in the target video.

Do the words or noises contradict the images? The audio may not appear to fit the person, particularly if the video was falsified but the original audio was not as meticulously modified.

Is the source trustworthy? Reverse image searching is a technique that journalists and scholars frequently use to determine the true source of an image, and it is something you can use right now. You should also look into who posted the photograph, where they posted it, and whether it makes sense for them to do so.

What do you think about this? Tell us in the comments.

Header image/Facebook image credit: AP

For more trending stories, follow us on Telegram.