LinkedIn says it targets generic-sounding posts rather than AI use itself, though it hasn't spelled out the exact triggers.

This explores what makes LinkedIn decide a post is inauthentic AI output and push it down in the feed. The short answer from the corpus: LinkedIn has described its targets but not the exact triggers. Its stated target is not "AI was used." It is posts that lack a clear perspective or voice. The company says these "generic" posts are kept mostly inside the author's own network instead of being shown more widely Does LinkedIn's generic content filter actually work fairly?. The reason it gives is that such posts dilute the conversations the platform depends on Does LinkedIn's AI detection actually improve conversation quality?. So the trigger is about sounding generic, not about provenance. A heavily AI-assisted post with a sharp personal angle could get through, and a bland human-written post could get caught.

The system appears to draw on two signals. The first is an automated classifier that LinkedIn says is 94% accurate. The second is people: a "Seems like AI slop" option in each post's three-dot menu Will LinkedIn's AI slop flag reduce AI-generated posts?. Those reports are meant to train the classifiers Does LinkedIn's AI slop button actually reduce low-quality content?. In practice, what readers find annoying becomes part of what the model learns to flag. LinkedIn has not published the features the classifier uses, how its training data was labeled, or its false-positive rate. That means nobody outside the company can say how often real writers get wrongly throttled How often does LinkedIn wrongly flag legitimate posts?.

The scale of the problem is large, though the numbers below come from outside detectors, not from LinkedIn. Originality.ai's detector labeled about 81% of a fixed sample of July 2026 LinkedIn posts as likely AI, up from about half in late 2024 How much LinkedIn content is AI-generated right now?. In an opt-in dataset analyzed by Pangram, LinkedIn accounted for most of the flagged AI content across platforms. Pangram suggests people lean on AI more when a post is tied to their real professional identity Why does LinkedIn have the most AI-generated posts?. Each of these tools measures its own definition of "AI," so none of them tells you what LinkedIn's own filter catches.

The corpus has one hint about why generic AI posts stand out. They tend to collect likes through polished, comprehensive phrasing but get few real replies, because nobody is arguing back Why do AI posts get likes without inviting conversation?. Over time that wears down the feed's job of building reputations for real human voices Does AI content displace human influencers on social media?. This lines up with LinkedIn's stated concern about conversation, but no note confirms that low reply rates feed into the filter.

The less obvious finding is that readers may enforce this more than the algorithm does. LinkedIn's CEO says AI post-writing tools get less use than expected because people who spot machine-written prose call it out publicly, and that costs the poster professionally Why aren't LinkedIn users adopting AI post-writing tools?. Separately, research on fake profiles shows how fragile detectors are. Detectors trained on older fakes missed roughly half of GPT-generated profiles, and only retraining on GPT examples fixed it Can fake profile detectors catch GPT-generated LinkedIn profiles?. Whatever LinkedIn's triggers are today, they probably only catch the writing styles they were trained on and will need updating as AI writing changes.