Shruti Kaushik

ResearchAI search

How Much of LinkedIn Is AI-Written? Reach and Trust Data (2026)

What the research says about AI-written LinkedIn posts in 2026: how common they are, how LinkedIn ranks them, and what disclosure does to reader trust.

By , , 5 min read

On this page
  1. 1. The share of AI-written posts: 54% in 2024, 81.2% in 2026
  2. 2. What LinkedIn says about reach
  3. 3. What happens to trust when readers know
  4. 4. Why buyers care about the substance
  5. 5. Detectors are a weak referee
  6. What this means for a founder
  7. Sources

The best available estimate says most long LinkedIn posts now carry the statistical marks of AI writing: 81.2% of a 5,000-post sample in July 2026, according to a detector vendor. LinkedIn now says generic, AI-looking posts get less distribution outside your network, and controlled experiments show readers trust a writer less once they learn AI was involved. What nobody has published is a clean, independent measure of how much reach a given AI-assisted post loses.

#1. The share of AI-written posts: 54% in 2024, 81.2% in 2026

Originality.ai's 2024 study ran 8,795 long-form posts (100 words or more) from January 2018 to October 2024 through its own detector and estimated 54% were AI-generated. It also found suspected AI use rose 189% from January to February 2023, right after ChatGPT's launch, and average long-form post length rose 107%.

In July 2026 it repeated the exercise. 4,061 of 5,000 public posts, or 81.2%, were classified as likely AI, using the same method.

Read both numbers with care. Originality.ai sells AI detection, and its write-up reports no false-positive rate. The sample comes from search results, not members' feeds. LinkedIn itself told The Register in 2024 that it does not track whether generative AI was used to create or edit posts, and described the flagged content as "AI-assisted" rather than fully machine-written.

#2. What LinkedIn says about reach

LinkedIn does not penalize AI use as such. In May 2026, Laura Lorenzetti, LinkedIn's VP and Executive Editor, wrote that content which appears AI-generated and lacks clear perspective "is less likely to be widely distributed beyond a person's immediate network." Automated comments at scale are targets too.

The line that matters for founders: "It's ok to use AI to help you write, but your posts and comments need to represent your voice and your perspectives." LinkedIn says its initial testing identifies generic content correctly 94% of the time. It has not published how often it wrongly flags a human post. Engadget reported that the company is also watching for stock AI phrasing such as the "it's not X, it's Y" construction.

In late July 2026, LinkedIn added a "Seems like AI slop" option next to "Not interested" in the post menu. Product chief Hari Srinivasan said the reports help tune LinkedIn's models and acknowledged automated detection can be wrong.

#3. What happens to trust when readers know

The academic evidence is consistent. Oliver Schilke and Martin Reimann ran thirteen experiments, published in Organizational Behavior and Human Decision Processes in 2025, and found people who disclose AI use are trusted less than people who do not. The cause was lower perceived legitimacy. One finding cuts the other way: being exposed by a third party did more damage than disclosing it yourself.

A second paper by Manav Raj, Justin Berg and Rob Seamans tested about 27,000 participants across 16 experiments. Telling readers AI was involved lowered evaluations of identical writing by 6.2% on average. Readers rated the work less authentic. Framing the AI as a collaborator did not reliably remove the penalty.

A 2025 preprint from researchers in Japan found the size of the disclosed AI share matters. Higher disclosed AI ratios produced significantly more negative shifts in perceived trustworthiness, competence and likability, with the sharpest drops for social, interpersonal writing. About 38.6% of responses were neutral or positive, mostly when the disclosed AI share was under half.

#4. Why buyers care about the substance

In the 2024 Edelman-LinkedIn B2B Thought Leadership Impact Report, a survey of 3,484 executives, 73% said an organization's thought leadership is a more trustworthy basis for judging its capabilities than its marketing materials. 60% said they would pay a premium to work with an organization that produces thought leadership. The report does not measure AI content. It shows buyers use this content to judge competence.

#5. Detectors are a weak referee

Every "percent AI" figure depends on a detector, and the peer-reviewed record on detectors is poor. A 2023 study in the International Journal for Educational Integrity tested 14 tools and concluded they were "neither accurate nor reliable", with paraphrasing making them worse. Stanford researchers found seven detectors classified 61.22% of TOEFL essays by non-native English writers as AI-generated. For founders writing in a second language, that bias is a real risk.

Finding Figure Source
Long-form posts flagged likely AI (2018 to Oct 2024) 54% of 8,795 Originality.ai via The Register
Posts flagged likely AI (July 2026) 81.2% of 5,000 Originality.ai
LinkedIn generic-content classifier, initial tests 94% correct LinkedIn
Drop in evaluations after AI disclosure 6.2% average Raj, Berg, Seamans
TOEFL essays misread as AI by 7 detectors 61.22% Stanford HAI
Execs who trust thought leadership over marketing 73% Edelman-LinkedIn 2024

#What this means for a founder

  1. Put one thing only you know in every post. A customer number, a decision you made, a mistake you paid for. LinkedIn's stated test is perspective, and generic posts are the ones losing out-of-network reach.
  2. Strip the stock phrasing before you publish. Cut "it's not X, it's Y" lines and recycled openers. LinkedIn has named these patterns.
  3. Do not rely on an AI detector to clear your drafts. The research says they misfire, especially on non-native English. Edit for specificity instead, and if you use a ghostwriter, make sure the ideas come from your calls and your data.

#Sources

By Shruti Kaushik

Quick questions

How much of LinkedIn is AI-written in 2026?

Originality.ai's July 2026 analysis classified 81.2% of 5,000 public LinkedIn posts of 100+ words as likely AI, up from 54% of 8,795 long-form posts in its 2024 study. These are detector estimates from the company that sells the detector, not a platform-wide count.

Does LinkedIn reduce the reach of AI-written posts?

LinkedIn says content that appears AI-generated and lacks clear perspective is less likely to be distributed beyond the author's immediate network. It also says using AI to help you write is fine if the post represents your own voice and perspective.

Does disclosing AI use hurt trust?

Yes, in controlled studies. Thirteen experiments by Schilke and Reimann (2025) found people who disclose AI use are trusted less, though being exposed by someone else hurts more. Raj, Berg and Seamans found disclosure lowered evaluations of writing by 6.2% on average across about 27,000 participants.

Keep reading