LinkedIn Also Shows AI Fatigue, Offers Tips on Writing Widely Read Posts

What happened

LinkedIn feeds have been filled with AI-made posts. These posts often use emoji lists, promote business deals, and ask people to follow or comment for more information. LinkedIn is now working to remove this kind of content. It also wants users to post in a more personal, human voice.

Key numbers

• A July report from the research firm Pangram found that over 40% of long posts on LinkedIn were made by AI.
• On X (formerly Twitter) and Medium, the share was about 30%. On Reddit and Substack, it was lower.
• In August, LinkedIn added a button that lets users report posts that look like AI slop. It was clicked more than 1 million times in its first two weeks.
• LinkedIn said last month that better detection cut views of content marked as AI slop by 40%.

How LinkedIn is responding

LinkedIn has trained its AI models to spot posts that seem real and useful. Posts flagged as low-quality AI content are less likely to be shared beyond the writer's direct connections. A team of former journalists and social media experts also coaches executives and experts. When posts do poorly, the team asks the writer whether the post sounds like them and whether they enjoyed reading it.

Why it matters

If users only see the same generic AI posts, they may leave for other apps. LinkedIn's content chief, Dan Roth, says real, personal posts work best. He also says AI can still help, but only if people edit it and add their own voice. Otherwise, posts are unlikely to reach many readers.

What to watch

• Whether the new report button and AI detection keep lowering views of AI slop.
• Whether users start posting more casual, unpolished videos and notes instead of polished AI text.
• Whether other platforms, such as YouTube, Pinterest, and Spotify, take similar steps against AI spam.