YouTube is pledging to crack down on low-quality AI content while simultaneously introducing AI tools that will let creators generate deepfakes of themselves.
In his annual letter published on the YouTube Blog, CEO Neal Mohan promised that YouTube will combat "AI slop" among other repetitive, clickbait, and spam content. He claims the company already possesses strong systems for detecting unwanted content, which it will keep building upon during the year.
Mohan mentions ASMR and gameplay videos as examples of once-unconventional content that are now widely accepted. The comparison implies the company may not have strictly defined guidelines and will likely approach content moderation on a case-by-case basis.
"The rise of AI has raised concerns about low-quality content, aka "AI slop." As an open platform, we allow for a broad range of free expression while ensuring YouTube remains a place where people feel good spending their time."
At the same time, YouTube announced plans to expand AI tools for creators, including a feature that lets them create Shorts using their own AI-generated likeness.
The company didn"t explain how this deepfake tool will work alongside its likeness-detection system introduced last October. The two systems will have to be compatible to avoid flagging creator-generated content as unauthorized deepfakes.
YouTube also revealed that over 1 million channels used its AI creation tools daily in December, with plans to let creators "produce games with a simple text prompt" and "experiment with music" in 2026.
Companies announcing new AI tools while simultaneously advocating for higher quality standards have become a common sight. Microsoft CEO Satya Nadella believes that the only way for AI not to turn into a bubble is for it to provide real-life value to users.
Zooming in, however, is more complicated, as the "AI slop" debate is still raging. Most companies are trying to find a balance until the pendulum swings into a clearer, more defined direction, and the public reaches a consensus on what"s acceptable as legitimate content and what"s considered "AI slop."