AI Slop Report: The Global Rise of Low-Quality AI Videos
Kapwing's "AI Slop Report" tries to quantify something most people already feel: it's getting harder to tell whether a random short-form clip is a real person, a low-effort repost, or a fully synthetic "content farm" optimized for the algorithm. Their headline estimate is that roughly 21-33% of a typical YouTube feed may be made up of what they call "AI slop" (high-volume, low-quality, often auto-generated video).
What makes the piece useful isn't the exact percentage - it's the framing of the incentive loop. Generative video lowers the cost of producing endless variations, while recommendation systems reward whatever drives clicks and watch-time, regardless of whether it's informative. That combination pushes platforms toward a moderation and attribution problem: when volume explodes, it becomes harder to enforce policies, harder for creators to stand out, and harder for viewers to develop intuition about what's authentic. If you're building with gen-video tools, it's a good reminder that "quality" is less about photorealism and more about intent, labeling, and distribution.