YouTube deleted 16 channels with a combined 35 million subscribers this year, wiping 4.7 billion views in one sweep. The platform is calling it a crackdown on "AI slop" - low-effort, mass-produced content that flooded the feed once generative video tools made it nearly free to produce.
At the same time, a different AI fight is playing out further up the chain. Creators are increasingly asking whether the platforms hosting their work are also training AI models on it, often without a clear answer, sometimes without being asked at all. The EU AI Act now requires disclosure of training data sources. Forbes called it the creator economy's "next war."
Two AI stories, same year, seemingly unrelated. They're not.
The slop problem
Mass-produced low-effort content isn't new. Before AI, it was content farms, clickbait thumbnails, and reaction-video mills - anything cheap to produce that an algorithm would reward with reach. Generative AI didn't create that incentive. It just made it dramatically cheaper to act on.
A platform that rewards upload volume and watch-time will always attract whatever is cheapest to produce at scale. For a while that was low-effort human content. Now it's AI. The tool changed. The incentive didn't.
Long Play has no algorithmic feed and no engagement-driven recommendation system to game. There's nothing here that rewards uploading more, uploading often, or optimising for watch-time. That doesn't mean AI-generated work isn't welcome. It is, and always has been. It means there was never a mechanism pushing you toward producing more of it than you actually wanted to make.
The ownership problem
The training-data fight is really about something older: who benefits from your work once you hand it to someone else's platform. AI training is just the newest way that question gets asked.
Long Play's answer hasn't changed since before this became an industry-wide fight. We will never use your videos to train AI models, full stop. Not for our own tools, not licensed to anyone else's. That's on the privacy page today, and it isn't a response to this year's news cycle. It's the same ownership principle behind everything else creators keep on Long Play, including the right to download your original files at any time.
Where AI actually shows up
To be clear, Long Play isn't an anti-AI platform. You can use AI in your work - generative visuals, synthesised voices, AI-assisted editing, whatever fits the piece - and all we ask is that you disclose it when it plays a significant role, so your audience knows what they're watching. We don't gate content by how much AI was involved. We trust you to be honest about it, the same way we trust you with everything else here.
Long Play itself barely touches AI at all. The one place it shows up is caption generation: an optional Whisper-based transcription that runs entirely on our own servers, produces a draft you have to review and approve before it ever reaches a viewer, and never sends your video anywhere else. That's it. That's the whole list.
Neither of this year's AI reckonings needed AI to happen in the first place. They needed an incentive to reward volume, or a policy that treated your work as raw material. Long Play was built without either, well before either became a headline.