Linkedin’s clampdown on AI-generated posts

LinkedIn is trying to stem the “AI slop”, the mass of generic and low-quality content produced with artificial intelligence that is invading the platform. The social media professional introduced last month the ability for users to report suspicious posts and comments through the “seems like AI slop” option, the Wall Street Journal reports.  The problem now concerns all the main social platforms, but for LinkedIn it takes on particular importance. The company, which has over a billion members, in fact, it bases much of its reputation on the quality of professional information, content and job opportunities shared by users.

According to an analysis by Pangram Labs cited by the American newspaper, 41% of long posts and 30% of comments published on LinkedIn between April and June were entirely generated by artificial intelligence. The research examined almost 57 thousand contents. The presence of AI would be higher on LinkedIn than on other text-based platforms: according to Pangram Labs, artificially generated content would represent 29% on X and 13% on Reddit.

Another company specializing in artificial content detection, Originality.ai, analyzed 5,000 public posts on LinkedIn in July and estimated that 81% had AI use above a level considered “moderate.” However, LinkedIn disputes these findings and claims that external analyzes tend to classify content simply assisted by artificial intelligence as “AI slop”. “AI can be a great support tool to help people articulate their ideas, refine language or make it more concise,” a company spokeswoman explained.

LinkedIn sells an AI writer. Now it collects slop reports. The button literally says slop, 404 Media found a candidate in about 6 seconds, and Pangram estimates 41% of long-form posts are likely AI-generated. https://t.co/JftT8QMx6b
pic.twitter.com/hMUgkpvoVm

— TNW (@thenextweb) July 31, 2026

Post signals made with AI

According to Dan Roth, editor in chief and vice president of content at LinkedIn, some signs are now easily recognizable: generic conclusions, unusual formatting, emojis used to solicit shares and comments, repetitive rhetorical constructions and long sequences of short, effective sentences. One of the most frequent clues, explains Roth, is the presence of well-written comments which however limit themselves to summarizing the content of the post to which they respond. LinkedIn has therefore decided to intervene on its own tools as well. Last month it announced the phasing out of the “enhance your post” feature, which allowed you to use artificial intelligence to correct and rewrite texts.

The decision, according to Roth, is linked both to low usage and to the desire not to further fuel the problem. The platform also plans to introduce a warning for users whose content is flagged by others as potentially AI-generated. At the beginning of the year it had already introduced an option that allows you to hide comments from unverified users.

LinkedIn blocks 200,000 spam comments a day

LinkedIn says it blocks more than 200,000 AI-generated spam comments before posting every day. The company also claims to correctly identify about 94% of generic or AI-produced posts, limiting their distribution beyond the author’s direct network. For LinkedIn, however, the point is not to determine whether or not content was created with artificial intelligence, but to evaluate its usefulness for users.

“How do we make sure that the time spent on LinkedIn is a good investment, that when you open the app you find something that makes you better at your job or at the job you would like to have,” Roth explained. However, accurately measuring the amount of content generated or assisted by artificial intelligence remains increasingly difficult. “A significant share of LinkedIn content is certainly assisted by AI, but no one can know exactly how much, not even LinkedIn,” noted Jérémy Boissinot, CEO and co-founder of Favikon.