LinkedIn Is Quietly Suppressing AI-Generated Posts — Here’s What the Data Shows

LinkedIn is actively suppressing AI-generated posts. Fully AI-written content receives approximately 2.8x less reach and nearly 5x less engagement than human-written posts, according to Richard van der Blom’s Algorithm Insights 2025 Report, built on more than 1.8 million posts across 400,000-plus profiles. The platform’s suppression operates through two overlapping systems: a 150-billion-parameter AI model called 360Brew, deployed in the live feed on March 12, 2026, and a formal editorial policy announced on May 20, 2026.

What LinkedIn Actually Announced

On May 20, 2026, LinkedIn Global Editorial VP Laura Lorenzetti published a post confirming three new anti-slop measures: reduced algorithmic reach for posts appearing heavily AI-generated without original perspective; stronger automated detection of AI-written comments; and new user filters to restrict feeds to verified profiles only. The announcement was explicit — as Social Media Today reported, Lorenzetti stated that AI overused at scale “dilutes the valuable insights that real human conversations can spark.” Flagged posts are not deleted. They are confined primarily to the author’s first-degree network, cutting off broader distribution without any visible warning to the creator.

How 360Brew Detects and Penalizes AI Content

360Brew, LinkedIn’s unified feed-ranking model built on LLaMA 3 and described in a January 2025 research paper, handles more than 30 predictive tasks simultaneously including feed ranking, ad targeting, and connection suggestions. Its AI-content signals are behavioral, not just textual. The model evaluates:

• Dwell time — how long a reader actually reads before scrolling past
• Save rate — whether anyone bookmarks the post for later
• Comment depth — whether replies are substantive or one-liners
• Lexical diversity and phrase repetition — flagging templated openers like “In today’s fast-paced world” and bullet-heavy structure with no personal voice
• Profile coherence — whether the post’s topic matches the author’s documented expertise and prior content history

LinkedIn claims 94% accuracy in detecting generic AI content, though the platform has not disclosed false-positive rates. SocialNexis notes that 360Brew evaluates “lexical diversity, tone consistency, phrase repetition, and expertise match” against an account’s full posting history — meaning the penalty compounds over time for repeat AI-volume posters.

The Reach Numbers by the Data

Multiple independent datasets now corroborate the suppression effect:

• 2.8x less reach, 5x less engagement for fully AI-generated posts vs. human-written (van der Blom, 1.8M post dataset, 2025)
• 70% peak-to-trough reach decline for organic content between February 2024 and August 2025 (Apostle Social, citing Algorithm Insights Report 2026)
• 45% lower average engagement for likely-AI posts in Originality.ai’s dataset of 3,368 long-form posts from 99 influential LinkedIn creators
• 73% more engagement per post for human-written content in marketing and branding, even though 61% of that category’s posts were likely AI-generated
• More than 80% of long-form LinkedIn posts sampled in July 2026 were likely AI-written (Vynce Digital, citing Pangram Labs data)
• Company page organic posts now represent approximately 2% of what appears in users’ feeds (Whitehat 2026 analysis of Socialinsider data)

The Dual Mechanism: Detection Plus Behavior

There is an important nuance that most reporting misses. 360Brew does not rely solely on text detection to suppress a post. Even a post that passes NLP classifiers will lose reach if readers scroll past it in seconds. The suppression comes from two directions simultaneously: NLP classifiers flag predictable AI patterns before distribution, and behavioral signals (near-zero dwell, no saves, no real comments) cut distribution after the initial push. A post can fail either test independently. This is why some AI-assisted posts that carry specific expertise and earn substantive comments still perform well, while generic human-written posts also get suppressed.

Understanding where you sit in LinkedIn’s content hierarchy matters for the same reason it matters in software pricing — your audience segment and perceived authority level determine what distribution you’re eligible for. The underlying logic is explored in Your Software Bill Is Decided by Which Customer Class You Fall Into: platform operators sort users by signals of value, and AI-slop posters are being sorted into the lowest distribution tier.

What Still Works

LinkedIn’s May 2026 announcement explicitly stated that AI-assisted content remains welcome if it contains original perspective, expertise, or meaningful contribution. The practical line: AI drafts that are humanized with specific results, first-person experience, and varied sentence rhythm avoid the suppression signals. Three posts per week with a real opinion now consistently out-distribute seven generic ones, according to ViralBrain’s June 2026 analysis. Saves and substantive comments — not reaction counts — are the primary distribution amplifiers under 360Brew.

Sources: Richard van der Blom Algorithm Insights 2025 Report (socialpilot.co summary, August 2026 — https://www.socialpilot.co/blog/linkedin-algorithm); LinkedIn Engineering Blog, Hristo Danchev, March 12 2026 (via zoomsphere.com — https://www.zoomsphere.com/blog/linkedin-algorithm-2026-why-generic-ai-content-kills-your-organic-reach); Laura Lorenzetti LinkedIn announcement, May 20 2026 (via Social Media Today — https://www.socialmediatoday.com/news/linkedin-wants-to-limit-the-reach-of-ai-generated-content/820935/); Originality.ai engagement analysis, January 22 2026 (via vyncedigital.com — https://vyncedigital.com/blog/linkedin-ai-content); Pangram Labs study (via TechTimes, July 15 2026 — https://www.techtimes.com/articles/320534/20260715/linkedin-leads-all-platforms-ai-written-posts-study-million-feeds-confirms.htm); SocialNexis AI engagement decay analysis (https://socialnexis.com/guides/ai-posts-engagement-decay-rate); Neil Patel Blog, LinkedIn AI Slop Crackdown (https://neilpatel.com/blog/linkedin-ai-slop-crackdown-content-strategy/)