LinkedIn reduces visibility of generic AI-generated posts by 40%

LinkedIn has implemented new measures to reduce the reach of posts identified as generic, repetitive, or low-quality content generated by artificial intelligence. The company reported a 40% drop in views for such content over recent weeks as a result of strengthened automated filters.

To identify this content, which the platform refers to as ‘AI slop’, LinkedIn utilizes large language models trained on user reports alongside rapid detection systems designed to spot repetitive structures. These filters target automated texts that lack original insight or personal perspective, limiting their distribution beyond the author's immediate connections.

Tim Jurka, Vice President of Engineering, stated that the expansion of these content classifiers covers all posts distributed beyond a user's direct contacts, boasting a 94% accuracy rate in identifying low-quality text. LinkedIn's strategy aims to encourage authenticity and human contribution rather than banning AI tools entirely, emphasizing that users should use AI as a support for writing rather than a tool for publishing hollow content.