LinkedIn has introduced a button that lets users flag posts and comments that “seem like AI slop”: low-grade, machine-generated filler. It is the first measure of its kind on the professional network owned by Microsoft, and it would be almost admirable if it were not coming from the same platform that, until recently, offered its users an AI-powered “improve your post” feature. The dealer has opened a rehab clinic, and the waiting room is full of his own customers. To be precise about what the button does: it reports a suspicion, it does not delete anything by itself, and a detector’s verdict is a probability, not a proof.
The numbers explain the urgency. Pangram Labs, a firm specialized in detecting machine-generated text, estimated that in June 41% of long posts and 30% of comments published on LinkedIn were produced with AI, against 29% on X and 13% on Reddit. Estimates from a detector, not court rulings, but the direction is hard to miss: the network built on professional reputation appears more saturated with synthetic prose than the platforms people join to argue about video games. Which makes a grim kind of sense: nowhere else is the incentive to appear brilliant so high and the willingness to actually sit down and think so low.
The pattern, when it is there, is familiar. A punchy opening engineered to stop the scroll. An assertive, oracular tone. A body made of bullet points, because paragraphs require transitions and transitions require thought. A closing question designed to farm comments. The comments follow the same template: chatbot formulas, recycled structures, the same ornamental emojis. An assembly line of frozen wisdom, where the same identical thought circulates under different names.
LinkedIn’s vice president of content describes a grey zone: posts that seem written by humans but are not entirely so. And the backlash writes itself. Users protest that they use AI to write better, not to deceive anyone. The distinction deserves to be taken seriously, because it is the whole point: a machine that fixes your grammar is assistance; a machine that thinks in your place is delegation. The target is not the professional who gets their subjunctives corrected. The target is the mechanism that rewards publishing, every single day, opinions nobody ever actually had. Grammar can be fixed in a minute; an opinion requires the inconvenient step of having one.
The contradiction, of course, has not gone unnoticed. The platform offered AI tools to improve posts, then handed those same users a button to denounce the results. This is not a change of heart; it is damage control. When an external detector estimates that four long posts in ten are synthetic, trust starts to evaporate, and trust is the only thing a professional network actually sells. Europe’s AI Act, with its transparency obligations for AI-generated content, hovers in the background of the same conversation.
The interesting part is the endgame. When everyone plays the same synthetic score, authenticity becomes scarce, and scarcity creates value: a human voice could become a visibility factor rather than a quaint habit. On LinkedIn, in 2026, the most ruthless growth strategy available may simply be writing your own posts. The machines flooded the platform, and the only disruptive move left is being a person.
Raffaele Di Marzio
All my “insane” books on cybersecurity and governance are here 👇 https://cyberium.limited/bookshelf.html
