YouTube used AI to edit videos

After age verification , YouTube introduced another new feature that sparked widespread protests. Without notifying users or asking for their permission, the Mountain View company launched a test using artificial intelligence to improve video quality. YouTube confirmed that it was an experiment.

It's not generative AI, but just machine learning

YouTube recently announced it would block the monetization of AI-generated videos. In mid-August, a popular YouTuber (Rhett Shull) posted a video demonstrating the strange differences between the version of the video on Instagram and the one shared with YouTube Shorts.

The YouTuber highlighted the flaws introduced by artificial intelligence on YouTube. According to a professional photographer, the video is fake (not authentic, to use YouTube's terminology). It essentially looks like a low-quality AI deepfake . According to Shull, it's a case of non-consensual AI upscaling.

A YouTube employee (Rene Ritchie) wrote on X that no upscaling was done and no generative AI was used. It's simply an experiment that uses traditional machine learning technology to remove blur, noise, and sharpen videos during processing, similar to the features available on a modern smartphone (computational photography).

The Californian company's goal is therefore to improve the quality of videos uploaded to the platform. The problem is that content creators were not notified , and there is no option to opt out of the experiment. As the YouTuber pointed out, unauthorized video editing could cause damage to their reputation. If users notice altered videos, they abandon their channel, and the content creator loses any earnings.