LinkedIn's AI slop button reaches one million reports

What this is about

LinkedIn introduced a feedback option labeled “Seems like AI slop” at the end of July 2026. Users can flag posts that look like automatically generated filler. According to a public post by LinkedIn chief product officer Tomer Cohen, the button was used more than one million times within a few weeks. Combined with updated detection, it cut views of content LinkedIn classifies as AI slop by 40 percent, the company says.

The figures matter because LinkedIn benefited for years from making writing easier: first templates, then writing assistants, and finally complete posts at the press of a button. The platform is now visibly addressing the side effect: more text but less genuine information.

What LinkedIn's feedback option actually does

The option does not place a public label under a post. A click gives LinkedIn another signal for ranking the feed. The platform combines user feedback with its own detection systems and can reduce the reach of suspicious content. LinkedIn does not say that every reported post is automatically removed, or that every piece created with AI is unwanted.

The important distinction is between the tool and the result. A person can use AI for spelling or structure and still publish an original, useful post. The feedback option targets content that appears interchangeable, mass-produced or misleading. LinkedIn does not disclose how much weight one report carries or which threshold triggers demotion.

Why it matters

A professional network depends on users being able to assess people, experience and expertise. When generic success stories, fabricated experiences or automated comments fill the feed, the cost of finding useful information rises for everyone. The million reports are therefore less a curious counter than a sign of a trust problem.

A July 2026 study by detection provider Pangram classified 41 percent of the longer LinkedIn posts in its sample as fully AI-generated. That is not a complete measurement of the entire network: sampling, language and detection methods affect the result. It does, however, fit the observation that LinkedIn is now explicitly relying on user feedback alongside technical filters.

For authors, the incentive is clear: original examples, verifiable numbers and concrete experience become more valuable than polished but empty prose. For companies, it also means automated publishing schedules can lose reach even when they do not formally violate a platform rule.

In plain language

Imagine the LinkedIn feed as a buffet. When more and more trays contain the same tasteless instant soup, a note from the guests helps the chef move those trays to the back. The note does not prove who cooked the soup. It only says that the offering feels interchangeable and is crowding out better food.

A practical example

A recruitment consultancy publishes 50 automatically generated posts per week. Every text contains the same general claims about leadership and success, with no names, data or first-hand cases. Ten users flag several posts as AI slop. LinkedIn's system also detects recurring phrases and reduces distribution.

Another consultant publishes one post per week about a real placement: 37 applications, three interview rounds and one specific lesson from the process. The consultant uses AI only to shorten the draft. The new feedback option does not evaluate the writing process; it gathers signals about perceived quality. That does not guarantee that the second post will receive greater distribution.

Scope and limits

• One million clicks do not equal one million conclusively proven AI posts. Repeat reports and incorrect judgments are possible.
• The reported 40 percent reduction applies to content detected by LinkedIn. Without a baseline, time window and testing method, the effect cannot be independently reproduced.
• AI detection can incorrectly flag human writing and miss edited AI text. The feature is therefore a ranking signal, not reliable proof of authorship.

LinkedIn also provides no complete definition of “AI slop.” Transparency about appeals and the influence of individual reports remains important so that coordinated campaigns or simple dislike do not produce invisible penalties.

SEO & GEO keywords

LinkedIn, AI slop, generative AI, feed quality, content moderation, Tomer Cohen, Pangram, AI detection, social networks, automated content