LinkedIn details AI-slop crackdown after a million users flag posts

LinkedIn is using former journalists and social media experts to train AI that spots low-quality, AI-generated posts and limits their reach. In an October 18 CNN interview, Editor-in-Chief Dan Roth explained the approach behind a crackdown whose reporting button drew more than a million users in its first two weeks.

The team helps teach LinkedIn’s model what authentic, useful posts look like. When the model flags a post as low-quality AI-generated content, LinkedIn makes it less likely to spread beyond the author’s direct connections. The consequence is reduced distribution, rather than simply a notice to the author.

The early numbers measure views, not removals

The initial results predate Roth’s interview. In August, Chief Product Officer Hari Srinivasan said more than one million people had clicked the Seems like AI slop option within two weeks. He attributed a 40% reduction in views of content LinkedIn classifies as AI slop to member reports and changes to its detection systems.

That company-reported figure measures exposure to content under LinkedIn’s own classification. It is not a count of posts removed or a measure of the reporting button’s effect alone. Srinivasan said LinkedIn combines multiple signals and has safeguards intended to prevent individual feedback from unfairly targeting other members.

Importantly, no single piece of feedback determines how content is distributed

Hari Srinivasan, LinkedIn chief product officer

Authors get feedback—and coaching

Alongside those early results, LinkedIn began adding messages to Post Analytics when posts received enough community feedback identifying them as AI slop. Srinivasan described the goal as helpful feedback and said the company approached the feature assuming good intent.

Roth also described a creator management team that coaches executives, subject matter experts and other users whose polished AI-written posts perform poorly. Its advice includes asking whether authors enjoy reading their own posts and recognize themselves in the writing.

The team encourages less polished contributions, such as informal phone videos or short posts meant to start a conversation. LinkedIn has also recently scaled back its AI editing and rewriting tools. Roth said the platform is not opposed to AI assistance, but content that users fail to edit or personalize is less likely to reach a wide audience.