More than 40 percent of longform posts on LinkedIn have been flagged as fully AI-generated, according to a study by Pangram, an AI detection company. That number alone tells you something about the scale of the problem. But what makes this week’s news actually interesting is that LinkedIn isn’t just pointing fingers at users. It’s also pulling back its own AI writing tools that contributed to the flood in the first place.
As reported by Engadget, LinkedIn is testing a new reporting feature that lets users flag posts as “seems like AI slop.” When flagged, the post’s author gets a private notification through the platform’s analytics dashboard that others found the content inauthentic or AI-heavy. The flagged posts also get reduced algorithmic reach, similar to how the “not interested” button already works. It’s a soft penalty, not a hard removal.
LinkedIn’s Chief Product Officer Hari Srinivasan said the company has also been improving its detection systems in the background, which should reduce how often AI-generated posts get recommended in user feeds. And the platform has been busy on the bot front too. Srinivasan said LinkedIn has caught hundreds of thousands of automated comments and blocked billions of automated posting attempts in just the last few months. That’s not a content quality problem. That’s closer to an infestation.
The more telling move is what LinkedIn is doing with its own tools. The platform is removing the “rewrite with AI” button that used to appear on posts and messages, which helped users wholesale replace their writing with LLM-generated text. In its place, LinkedIn is introducing a lighter-touch proofreading feature that corrects errors without changing the author’s voice. That’s a meaningful shift in philosophy, and an implicit acknowledgment that the earlier feature made things worse.
For context, LinkedIn isn’t the only platform wrestling with AI content at scale. X, Facebook, and Medium have all seen similar complaints. But LinkedIn’s professional identity context makes the problem hit differently. When someone’s career credibility is tied to what they post, AI-generated thought leadership is a trust issue, not just a content quality issue.
Whether these changes are enough is a fair question. Crowdsourced flagging systems are easy to abuse or ignore, and users who want to post AI slop can still do it. But combining user flagging with algorithmic demotion and pulling the platform’s own AI writing tools is a more coherent strategy than most social platforms have tried. It’s a start, even if it’s a late one.
