Best ways to spot AI deepfakes in photos and videos

AI deepfakes have heightened concerns about online safety as various celebrities have been featured in phony videos and pictures. Deepfake videos are clips where an individual's body and face are digitally distorted to look like someone else. With the advancement of technology, it is now possible to create imperceptible deepfake videos or images of any person.

This matter is a genuine concern among the public, who are now eager to quickly find any unusual signs of a deepfake video or image. This article takes a look at some things to look out for if one wants to detect AI deepfakes in photos and videos.

Note: Some aspects of this article are subjective and solely reflect the opinions of the writer

---

What are AI deepfakes, and how are they made?

Utilizing intricate algorithms and advanced machine learning techniques, deepfake technology can create and manipulate video content, producing convincing depictions of individuals doing or saying things that are entirely fabricated.

The Generative Adversarial Network (GAN), a combination of encoder and decoder networks, is used to produce deepfake videos. The encoder takes the source content and extracts important features and representations from it, such as the face in the original video. Afterward, these features are given to the decoder network to create new and manipulated content, like a modified face.

In the creation of video deepfakes, consistency is of utmost importance. To achieve this, the AI goes through a repeated process until the desired result is reached, and this is done for each frame of the video.

---

How to detect if a video or photo is AI-generated

One must pay attention to visual and audio inconsistencies, along with other suggestive signs, to identify AI deepfakes in both photos and videos.

The following factors, in particular, should be kept in mind:

• Facial expressions: Facial expressions that appear unusual should be highlighted, along with lip movements that fail to sync properly. Additionally, sporadic blinking may be indicative of something out of the ordinary.
• Background and vicinity: In a video, analyze if the lighting is uneven or if there are any visual aberrations. Make sure the person in the video could conceivably be present in that particular location.
• Source: Suspicions may arise if a random social media account is the origin of the media. To ensure its authenticity, you should double-check the source for reliability and reputation.
• Sound quality: To detect authenticity in a video, pay close attention to irregularities such as unnatural speech patterns or pitch or tone fluctuations.

Additionally, consider the plausibility of the individual's location in the video. It is important to analyze if their actions and words coincide with their well-known traits and behaviors.

Technology is becoming more sophisticated, which means it's getting harder to detect AI deepfakes. To tackle this issue, it's crucial to come up with improved detection methods while still using AI ethically.

With technology constantly evolving, it becomes necessary to be careful and watchful when consuming media.