July 31, 2026, (Inside AI) — LinkedIn is now letting users flag AI-generated slop directly in their feeds, adding a 'seems like AI slop' reporting option as part of a broader push to clean up low-quality automated content on the professional network.
The feature, announced Thursday by chief product officer Hari Srinivasan, arrives alongside new detection classifiers and a pivot away from AI writing enhancements. The move targets the flood of formulaic posts that have turned LinkedIn into what one analysis called the most AI-saturated platform.
According to a Pangram Labs report, over 40% of long posts on LinkedIn are entirely AI-generated. That figure underscores why the company is treating AI slop as a top priority, even as it acknowledges the difficulty of defining exactly what qualifies.
"AI slop is a top priority for all of us. We really care about this. People come to LinkedIn to connect with real people and share their real perspectives, ideas and expertise. Here are a few more changes to keep it that way," said Hari Srinivasan, chief product officer, LinkedIn.
The reporting button is the most visible part of a multi-layered defense. Srinivasan said LinkedIn has been ramping up automated defenses, catching hundreds of thousands of automated comment attempts daily and blocking billions of other automation attempts in recent months. New classifiers will identify AI slop and low-quality content, reducing its visibility in suggested feeds and outside-network content.
LinkedIn is also testing a private flagging system in user dashboards, letting members signal when their own posts may feel inauthentic or overly reliant on AI. The approach reflects a key insight: AI and slop are not synonymous, and human judgment remains essential.
Srinivasan explained that user research revealed many people turn to AI writing tools because LinkedIn's emphasis on detailed professional sharing makes them self-conscious about their writing. In response, the platform is removing its 'Enhance your post' feature and replacing it with a proofreading tool that corrects grammar and spelling without altering intent or style.
Why Detection Alone Won't Fix the Slop Problem #
Automated detection has limits. Research from Stanford's RegLab shows that AI text detectors remain unreliable, especially for non-native English writers. LinkedIn's decision to combine user reports with backend classifiers, rather than relying solely on detection scores, sidesteps some of those accuracy pitfalls.
Still, the platform faces a fundamental tension. Its algorithmic feed rewards frequent posting, which incentivizes AI-assisted content. Even as LinkedIn cracks down on slop, it must avoid penalizing legitimate AI use, such as accessibility tools or translation aids. The company's emphasis on human feedback loops suggests it is trying to thread that needle.
Other platforms have struggled with similar challenges. Meta and X have both experimented with AI content labels, but enforcement remains inconsistent. LinkedIn's professional context may give it more leverage to demand authenticity, but the sheer volume of AI-generated text makes moderation a moving target.
LinkedIn's Proofreading Pivot Signals a Broader Shift #
The retirement of 'Enhance your post' marks a strategic retreat from generative AI writing features. By offering only proofreading, LinkedIn is betting that users want help with mechanics, not substance. It's a notable departure from competitors like Grammarly or Jasper, which push AI-powered rewriting.
Srinivasan framed the change as a response to user feedback: people wanted to sound more professional, not have their voice replaced. The proofreading tool is expected to roll out gradually, with the company monitoring whether it reduces slop without discouraging posting.
The move also aligns with broader industry soul-searching about AI's role in communication. As generative models become embedded in writing tools, platforms are being forced to decide where assistance ends and inauthenticity begins. LinkedIn's answer, for now, is to draw a hard line at stylistic intervention.
Meanwhile, the platform continues to invest in automated defenses. Srinivasan noted that comment-level detection alone is catching hundreds of thousands of automated attempts each day, a figure that hints at the scale of the problem. Whether user reporting can meaningfully augment those systems remains an open question, but the new button gives members a direct stake in shaping their feeds.