Does AI-written content hurt your reach on LinkedIn?

The feed is majority machine now

The best measurement we have comes from Originality.ai, which ran its detector across 3,368 long-form posts (100+ words) from 99 influential LinkedIn profiles spanning 11 industries. Across January to November 2025, 53.7% of those posts scored as likely AI-generated. Their earlier study tracked the same shift structurally: the share of likely-AI long posts jumped over 50% after ChatGPT launched and never came back down. By mid-2026, coverage across the industry press describes LinkedIn as the platform where AI-written text is most dominant.

One caveat we hold ourselves to: AI detection is probabilistic. These are strong, directionally consistent estimates, not exact counts.

What the engagement data actually shows

Here is the nuance most takes skip. The same Originality.ai analysis found AI content does not uniformly underperform. In generic motivational categories, likely-AI posts actually pulled more engagement than human ones. But the pattern flips hard where trust is on the line: in categories like healthcare and government, human-written posts drew roughly 40 to 44% more engagement than likely-AI posts, and follow-up coverage of the data lands on the same conclusion: authenticity wins where credibility matters.

Founder content lives on the trust-dependent side of that split. Decision-makers explicitly use content as a capability signal: in the Edelman-LinkedIn research, 73% say thought leadership is a more trustworthy basis for assessing a company than its marketing. A post that reads machine-made under a founder's name is graded on exactly that axis.

Is there an algorithmic AI penalty?

We could not verify a blanket penalty for AI-assisted writing, and claims of a secret switch do not trace to anything public, so we won't repeat them. The documented dynamics are simpler: the feed is flooded with posts sharing the same constructions (the reversal cliches, the em-dash cadence, the engagement-bait closers), platforms visibly push against low-quality bait patterns, and readers have learned the tells. When 53.7% of long posts are likely AI, the penalty is competitive, not conspiratorial: sameness is invisible.

The working approach

The founders winning in this environment are not abstaining from AI. They are constraining it: real stories, real numbers, and their own phrasing in, drafts scanned against the known AI tells before anything ships, and nothing published that they would not say out loud. AI as leverage on a real voice, never as a replacement for one.

Slingapult's read: this is why our composer has a blocking anti-slop gate. Drafts are written from your own posts and material, scanned against the tells that mark text as machine-made, and rewritten until they read human, because your content is a trust asset before it is a reach asset.