The complaint dates back to June, and shorts users began sharing close-ups of their strange-looking face and body parts
Google has admitted to secretly modifying videos uploaded to YouTube shorts with AI, drawing criticism from creators who say their work was manipulated without consent.
For months, some YouTubers have noticed something odd about their videos. The face, clothing and background look subtle – sharp wrinkles, distorted ears, smooth skin, more uneasy, skin that appears more clearly in an almost plastic way.
The truth is now coming out. Google is secretly testing the video enhancement features of YouTube shorts, which use AI to change the creator's work. However, the rollout was done without providing the option to notify the uploader or opt out.
Rick Beato, a music YouTuber with over 5 million subscribers, expressed his surprise to the BBC, saying that his own face “looks strange.”
“I was like, 'Man, my hair looks strange,'” he said. “And as I saw, I almost looked at it.
Beat had largely dismissed the strangeness as an imagination until fellow creator Rhett Shall noticed a similar distortion. The video revealing Shull's results has already won over 500,000 views.
“If I had wanted this terrible, excessive trembling, I would have done it myself,” commented Shall.
“But the bigger thing is that it appears to be being generated. I think it's deeply misrepresenting my voice on the internet, and it can erode trust my audience in small ways.
The complaints date back to June, with shorts users beginning to share close-ups of strange faces and body parts. Speculation quickly swirled online about hidden AI operations.
YouTube Editor-in-Chief Rene Ritchie writes that video enhancements are not generative AI, but rather uses “traditional machine learning” techniques to scrape images, reduce blurring, and reduce noise, just like the latest smartphone video recording techniques.
He added that YouTubee It's about taking creator feedback seriously and “working on how to deliver the best video quality and experience.”
Despite this explanation, criticism remains intense.
Samuel Woolley, chairman of Dietrich's disinformation research at the University of Pittsburgh, argued that YouTube's language is misleading, saying “machine learning is actually a subfield of artificial intelligence.”
Woolly believes the company is underestimating AI use due to widespread concern. More importantly, he denies YouTube for changing the creator's work without transparency or consent.
Jill Walker Letberg, a professor at the Digital Narrative Center at the University of Bergen, warned that the controversy reflects a broader shift in how reality is mediated and manipulated through technology.
This incident follows an awkward pattern of AI reshaping the media without full disclosure. Earlier this year, Netflix faced a backlash from AI “remastered” of 1980s comedy like Cosby Show, creating surreal distortions and glitches.
Similarly, Samsung has been caught enhancing its moon photos with AI, and Google Pixel devices already offer AI-driven “Best Take” features, sewing photos together to create an artificial smile.
Google's new Pixel 10 encompasses the AI to generate extraordinary 100x zoom images, far beyond the limits of physical hardware.
