YouTube quietly lets AI edit the videos you upload

For months, YouTubers have been quietly complaining that something is wrong with their recent uploads. After a popular music channel dug deeper, Google has now confirmed that it is testing a feature that uses AI to enhance video quality.

The company claims this is part of an effort to “deliver the best video quality,” but it’s odd that they’re starting the test without notifying creators or providing any way to opt out of the test.

Google's AI makes YouTube videos look artificially sharp, making viewers think the content isn't real. Photo: Rhett Schull

Google's experiment caused a stir when it first rolled out to YouTube Shorts earlier this year. Users reported strange artifacts, distorted edges, and distracting smoothness that made viewers feel like they were edited with AI.

If you've ever taken a screenshot and zoomed in on a photo after taking it with your smartphone and noticed everything looked too sharp or like an oil painting, that's the result of Google's video processing AI.

According to Rene Ritchie, YouTube's Editor-in-Chief, this feature isn't quite like the AI features Google has been cramming into every other product.

In a post on X (formerly Twitter), Ritchie said the feature isn't based on generative AI, but instead uses "traditional machine learning" to reduce blur and noise while sharpening images. That's a subtle distinction, though—it's still a form of AI used for video editing.

YouTube forces content creators to label AI when using it, but does it silently without users' consent.

YouTuber Rhett Shull began investigating what was happening to his videos after discussing the issue with a fellow creator. He quickly realized that YouTube was applying AI video processing technology without informing anyone—he called it “upscaling,” though Google’s Ritchie insisted it wasn’t technically upscaling.

Whatever you call it, Google has finally admitted that it’s editing videos as part of an experiment. Whether this experiment will become an official part of YouTube’s upload process remains to be seen, and we also don’t know if the AI enhancement will be an option for creators.

Do you deserve to know?

Some casual viewers will probably love Google's AI-enhanced videos, but professional video makers are understandably annoyed.

Aside from the fact that the tweaked videos might look better (they don’t), Google is making these changes without giving anyone any notice. The company says it’s taking into account viewer and creator feedback to “iterate and improve” the upgrades.

Users need to know and decide whether to let AI interfere with their content.

When Google announced the Pixel 10 phones, they noted that their improved image processing could automatically AI-enhance photos.

The team believes it's important to be transparent about this, so the Pixel 10 has built-in C2PA labels. Photos you take can be automatically labeled as AI-edited to make sure everyone knows they may not accurately reflect reality.

But what about YouTube videos? The company seems less concerned about that. Though Google doesn’t like to admit it, users are reacting negatively to AI content.

If there’s any indication that a content creator is using AI, the internet will jump in. By implementing AI editing (or whatever Google wants to call it), the company could potentially subject content creators to undue scrutiny and potentially discredit its previous secret video testing.