techcrunch.com web signal

LinkedIn adds 'AI slop' report button, kills own AI writer

TL;DR

  • LinkedIn added a 'seems like AI slop' report button and new classifiers to filter low-quality generated posts from out-of-network recommendations.
  • The company says it is blocking hundreds of thousands of automated comment attempts daily and millions of other automation efforts in recent months.
  • LinkedIn is discontinuing its 'enhance your post' AI writer and replacing it with a proofreading tool that corrects grammar without altering voice.

There is something quietly telling about a professional network killing its own AI writing tool at the same moment it hands users a button to report AI slop. LinkedIn did both, TechCrunch reported, and the combination is more interesting than either move on its own.

The new option, labelled 'seems like AI slop,' sits alongside a set of classifiers meant to keep low-quality generated posts out of the recommendations users see from outside their networks. LinkedIn's chief product officer Hari Srinivasan told TechCrunch that 'AI slop is a top priority for all of us. We really care about this,' framing the work around the idea that 'people come to LinkedIn to connect with real people and share their real perspectives, ideas, and expertise.' The company says it is already blocking hundreds of thousands of automated comment attempts daily, and millions of other automation efforts in recent months. Users whose posts read as inauthentic will reportedly get a private nudge inside their dashboard rather than a public penalty, aimed at encouraging improvement rather than punishing assistance with refinement.

The self-reversal is the more revealing detail. LinkedIn is discontinuing its 'enhance your post' AI writing feature and replacing it with a proofreading tool that corrects grammar without altering the writer's voice. That is a platform tacitly admitting its own generation feature contributed to the sludge it now wants to sweep up, and a bet that authenticity signals are worth more to a professional network than engagement produced by a template.

The honest caveat is that a report button and a classifier are only as good as the labels behind them, and the reporting does not spell out how the system distinguishes AI-assisted refinement from AI-generated slop, how creators can appeal a private inauthenticity flag, or whether enforcement will bite as hard on high-follower accounts as it does on smaller ones.

Where this leads, if it works, is a version of the feed where human writing gets more oxygen than it has had for the last couple of years, and a template other professional feeds could copy while their own users tune generated posts out.

Shared on Bluesky by 2 AI experts