AI-generated content is saturating the information environment, and efforts to detect and remove it keep failing. A perspective in Nature Human Behaviour suggests looking at the problem from the other side: when people call AI content "slop", they are making a social judgement.
So and colleagues argue that slop is not a technical property of the content itself. It is what a particular community recognizes as unwanted AI content. The same generated text may be acceptable in one forum and dismissed as noise in another.
Framed this way, the authors say, the problem becomes tractable, measurable and governable. Researchers can study how community rules define acceptable AI content, which cues people use when they judge it, and how those judgements shift over time, instead of waiting for a perfect detector.
For behavioural scientists, it places a central question of platform governance within familiar territory: social norms, group identity and trust.
