YouTube using AI to tweak user videos without their knowledge

The way people experience reality is changing. Increasingly, what reaches us has already passed through an algorithm, shaped and smoothed by artificial intelligence, before we ever see it.

Now, some YouTube creators say the platform is applying these AI filters and tools to their content without their knowledge or consent.

Rick Beato notices subtle changes

In an interview with the BBC, music YouTuber Rick Beato said he began noticing little changes in his videos. His face and hair looked different, as if a filter had been applied, and he said it gave the appearance of wearing makeup. Beato, who has uploaded nearly 2,000 videos to his channel, said the change appeared in his most recent uploads.

AI tweaks without consent

YouTube has quietly used AI to tweak videos without telling creators. The changes are subtle, but some creators say it gives their content an unnatural, AI-made feel.

A Reddit thread first highlighted the changes. One user said their VHS-style videos were automatically altered by AI when uploaded, with added effects that completely changed the original look — and there was no way to disable the feature.

YouTube responds

After multiple complaints and a recent post to YouTube’s liaison on X, Rene Ritchie, the company confirmed it is testing features to enhance video clarity, similar to smartphone tools like unblur and denoise. Ritchie said they aren’t using generative AI or upscaling. 

“YouTube is always working on ways to provide the best video quality and experience possible, and will continue to take creator and viewer feedback into consideration as we iterate and improve on these features,” said Ritchie. 

He clarified his comments and terminology in a subsequent post about the experiment, following feedback that YouTube was being deceptive.

Ritchie said GenAI usually means newer technologies like large language models or transformers. Upscaling is the process of enhancing a lower-resolution video, such as 480p, to improve its quality in a higher resolution, like 1080p.

That’s not what’s happening here. Instead, YouTube is using the same type of machine learning found in smartphone cameras, like computational photography, to enhance videos without changing their resolution.

Creators voice concerns

Creator Rhett Shull posted a video raising concerns that he never agreed to have YouTube use AI to modify his content. His video has racked up more than 700,000 views. Shull said the unannounced changes undermine the trust and partnership creators have with the platform.

“Replacing or enhancing my work with some AI upscaling system not only erodes that trust with the audience, but it also erodes my trust in YouTube,” he said. 

How YouTube uses AI across the platform

YouTube relies on AI in a variety of ways that go beyond video recommendations. The platform uses machine learning to help moderate content, flagging videos that may violate policies before a human even reviews them.

AI also helps place ads, deciding when and where they’ll appear to keep viewers engaged while limiting interruptions. 

AI optimism divided: experts vs. the public

AI experts are far more positive than the public about the technology’s impact. In a Pew Research survey, over half of experts say AI will positively affect the U.S. over the next 20 years, compared with just 17% of adults. 

On personal impact, 76% of experts expect AI to benefit them, versus 24% of adults. When it comes to jobs, 73% of experts see a positive effect, compared with 23% of the public.