How to Block AI-Generated Videos on YouTube (What Actually Works)
Last updated: August 2026
Quick answer: Nothing detects all AI content reliably, and tools claiming otherwise overpromise. What works in practice: install the free Dehook extension, add AI-generated content as a topic to catch the obvious cases, and ban the repeat-offender channels with one click as you meet them. The combination shrinks slop dramatically even though perfect detection is impossible.
AI slop is the fastest-growing complaint about YouTube, and the hardest to filter honestly. Anyone promising a clean feed with one click is selling certainty that the technology cannot deliver.
What the layers catch
Dehook attacks from three angles: the topic filter catches formulaic titles, channel bans remove the factories one by one, and the length floor removes the volume-optimized output wholesale. None is perfect; together they turn a flood into a trickle.
Reduce AI slop in 3 steps
- Install Dehook. Add the free extension to your browser.
- Add the topic. Add AI-generated content as a blocked topic. Formulaic slop titles classify surprisingly well.
- Ban channels on sight. Slop comes from repeat offenders. The ban button on every video makes each encounter the last one.
Dehook↑ topic plus channel bans
The anti-slop combination in the Dehook popup.
FAQ
Why can nothing detect AI content reliably?
Because AI slop is a production style, not a topic: synthetic voices and generated imagery live in the audio and video, which title-based filters cannot see. Even dedicated detectors produce false positives and misses. Honest tools say so.
What share can filtering realistically catch?
The formulaic bulk: listicle farms, fake-story channels, and template titles classify well. The polished edge cases slip through, which is where per-channel banning covers the gap.
Does the video length filter help?
Yes, meaningfully. Slop economics favor high-volume short output, so a 4 minute floor removes a disproportionate share of it.
Will YouTube fix this natively?
YouTube requires disclosure for some synthetic content, but enforcement is spotty and disclosure does not remove anything from your feed. Client-side filtering is currently the only lever you control.