As nearly half of LinkedIn’s long posts are now AI-generated, platforms respond to rising ‘AI-slop’ with new detection tools and stricter rules to protect authenticity and maintain user trust.
Social media platforms are tightening their rules as AI-generated posts, images and videos flood feeds with material many users find repetitive, bland or misleading. On LinkedIn, the pressure is especially visible. The professional network has introduced a way for users to flag suspected low-quality AI posts, after research suggested that a large share of its longer posts are now machine-made.
That research, from Pangram Labs, found that almost half of LinkedIn posts longer than 250 words were fully generated by AI. The study analysed more than 1 million posts across LinkedIn, Substack, X, Medium and Reddit, and found that LinkedIn, while responsible for only about a third of the sampled content, accounted for 62% of all AI-flagged material. Pangram also found that X had the highest share when mixed human-AI posts were included, while Substack had the lowest. Similar findings have been echoed by TechRadar and The Decoder, which both reported that LinkedIn is the most saturated of the major platforms for long-form AI content.
The problem has a name: “AI-slop”, a phrase used for cheaply produced, low-effort AI content that can look polished but adds little substance. Marijn Sax, an AI researcher at the University of Amsterdam, told AD that such material is attractive to people who want to appear more credible than they are. Sax argued that it is also useful for online grifters, political extremists and others who want to flood the internet with fast, persuasive content without doing much work.
LinkedIn says the issue is a priority because users come to the platform to read what real people think, not machine-generated filler. Other platforms are also responding. According to recent reporting, Snapchat is downplaying AI videos, YouTube has moved to reduce or demonetise repetitive generative content and Substack has added a tool to detect whether a post was written with AI. Substack chief executive Chris Best has warned that systems which reward inauthentic material can push the entire ecosystem into a race to the bottom.
The broader concern is not only quality but trust. Pangram says people tend to trust AI-written material less than human writing, and complaints about “AI slop” are now common on Reddit and elsewhere. Sax said platforms may fear that users will eventually leave if their feeds are dominated by generic machine output. He also noted the practical limits of moderation: platforms can flag and reduce AI content, but they cannot reliably prove how every post, video or image was made, and outright bans would be difficult to enforce at scale.
Disclaimer: This content is intended for informational purposes only. Readers are advised to exercise their own judgement, conduct due diligence, or consult a qualified expert before acting on any information provided.





