A YouTube channel called “Joe Liza WWE” has been pumping out two-hour AI-generated wrestling videos where the robotic narrator audibly malfunctions, repeating words like “what” and “whoa” for minutes at a time before devolving into what one viewer described as obscene mouth noises. According to Futurism AI, the channel has uploaded roughly 90 of these videos in the past month alone, and the glitches are so extreme that commenters are openly mocking the host for “having a stroke four minutes in.”
This is significant because it confirms what creators and platforms have been quietly worrying about for months. Long-form AI content is now being published at scale without a single human ever watching it end-to-end before hitting upload.
What’s Actually Happening
Futurism AI reports that the channel was originally created in 2007, sat dormant for years, and recently came back to life as a slop factory. The output mixes real wrestling clips, AI-generated footage, and randomly inserted gameplay from the “WWE 2K” video game. The voiceover is the part that’s breaking.
A Bluesky user surfaced the broken uploads, and the comments section turned into a roast. One clip, since made private, had the narrator chanting “whoa” in escalating intensity until it stopped sounding human at all.
It’s not just weird audio. Futurism AI flagged that the channel is also pushing flat-out disinformation, including a video falsely claiming Chuck Norris was “killed” and another fabricating an arrest of wrestler Jade Cargill.
Why It Matters Now
YouTube has been playing what Futurism AI calls a “largely symbolic game of Whack-a-Mole” with AI slop. The platform’s monetization gate (1,000 subscribers or 4,000 public watch hours in 12 months) was supposed to filter low-effort spam. It isn’t.
The deeper issue is the economics. Generating a 2-hour video costs almost nothing. Even a fraction of a percent click-through from auto-play recommendations can make the math work, especially when channels stack 90 uploads in 30 days. Real creators competing for the same recommendation slots are getting buried by output they can’t match in volume.
A few channels have started branding their playlists as “no AI” to signal authenticity. That’s a market signal worth watching. When users have to be told a human made the video, the trust default has flipped.
What Practitioners Should Take From This
- If you’re a creator: Authenticity markers (face on camera, named host, verifiable expertise) now carry premium value. Lean into them.
- If you’re a platform operator: Watch-time metrics alone can’t filter slop anymore. AI generators optimize directly for those signals. Behavioral signals (comment sentiment, replay patterns, share-to-view ratios) hold up better.
- If you’re a brand buying ad inventory: Ask your DSP partners exactly how they’re identifying AI-generated channels. Pre-bid filters built before 2025 likely don’t catch the current wave.
- If you’re building AI content tools: The reputational ceiling for unmoderated long-form generation is collapsing in real time. Quality control isn’t optional.
The Bigger Pattern
Futurism AI notes this isn’t isolated. Last year, 404 Media exposed a “True Crime Case Files” channel running the same playbook with fabricated cases. Earlier reporting documented AI-generated children’s content flooding YouTube Kids. Each incident is treated as a one-off. They’re not.
What stands out here is the failure mode. The narrator doesn’t just sound robotic, it visibly breaks down inside the video, and nobody at the channel noticed or cared enough to fix it. That’s the tell. The operator never watched their own content. The platform’s recommendation engine did the watching for them, and apparently kept serving it up.
Expect platforms to start requiring AI-disclosure labels with real teeth, not voluntary checkboxes. Expect advertisers to start pulling spend from channels that can’t pass a humanity check. And expect the “made by humans” label to become a marketing category of its own.
Full details at the original source.