This gets talked about a lot on LinkedIn but the question is still surfacing on Reddit, frequently with tech executives and founders. We’re answering your questions about it here.

Answer: Only partly. A July 2026 study of over a million posts found LinkedIn has more fully AI-generated content than any other platform, and those AI-only posts get about 45% less engagement than human-written ones. Use AI to think and structure. Write the sentences yourself, or have someone who knows your story write them.

How much of LinkedIn is actually AI-generated right now?

More than you think, and probably more than you’d guess about your own feed. AI detection firm Pangram Labs analyzed over a million posts across 6 platforms in July 2026 and found that 40 to 41% of LinkedIn’s long-form posts, the ones over 250 words, were flagged as fully written by AI. That’s the highest share of any platform they studied, well ahead of Substack, X, and Reddit, where fully AI-generated long-form content stayed closer to 10 to 13%.

A separate study by Originality.ai, running the same month, put the number even higher using a broader “likely AI” threshold: Over 81% of a smaller 5,000-post sample. Different methodology, same direction. LinkedIn confirmed it is now rolling out a native flag members can use to report a post that “seems like AI slop,” and called cleaning it up a top priority.

If you post there and you’re funded, this matters. Investors and enterprise buyers check your LinkedIn before a call, same as they check your deck. What they find is starting to matter more than it used to.

Does it actually hurt you to post AI-written content?

On average, yes. Real numbers from Pangram: Fully AI-generated posts pulled about 45% less engagement than human-written ones. The only exception worth knowing is leadership and inspiration content, where AI-assisted posts actually outperformed by around 75%, likely because that category rewards a clean, structured take more than raw voice.

For a funded founder, the real risk is fabrication. I was reviewing an AI-drafted post with someone on my team recently, doing the normal polish pass, and it had inserted a specific claim about work that never happened. Confidently. In a tone that read like me. That’s what AI does when you ask it to fill in details you didn’t give it. Instead of saying “I don’t know,” it invents something plausible and moves on.

For a founder mid-raise or mid-enterprise-sale, that’s a fact an investor or buyer can check.

The tells are also getting easier to spot, by machines and by people. Em dashes are still the single biggest giveaway. Stacked “it’s not this, it’s that” contrasts are right behind them. Semicolons dropped into a sentence that’s supposed to sound like you talking are another tell.

How should a funded founder actually use AI on LinkedIn?

For structure and repurposing. Keep the sentences themselves human. Only about 5% of your audience sees any single post, so reusing a post that already worked is simple math, and I wrote more about what actually controls that reach in The LinkedIn Algorithm Actually Hates Your Executive Brand. The move that works: Keep the body of a strong post intact and have AI generate 2 or 3 new versions of just the hook and subhook, so the algorithm doesn’t flag it as a duplicate and suppress it.

What doesn’t work is asking AI to generate a full post from a 2-line prompt and publishing it as is. Read every AI-touched draft closely enough to catch anything it made up, beyond the em dashes.

The 3 Moves To Make This Week

Go back through your last 10 AI-assisted posts and scrub every em dash and every “it’s not this, it’s that” construction. Those are the 2 most obvious tells, to readers and now to LinkedIn’s own detection.

Read your next AI draft once specifically hunting for invented specifics: A number, a claim, an experience you didn’t actually have. AI will supply them confidently if you don’t.

Stop asking AI to write new posts from scratch. Take a post that already worked, keep the body, and have AI give you 2 or 3 new hooks and subhooks instead. That’s repurposing, and it’s the one AI use case that actually holds up.

Want 60 to 80 sales conversations started for you each month, with zero automation?

Apply to work together

Your friendly tech exec and founder LinkedIn guy,

Steve,
CEO
TORC

Frequently asked questions

How can I tell if a LinkedIn post was written entirely by AI? Look for em dashes, stacked “it’s not this, it’s that” contrasts, and semicolons dropped into casual sentences. LinkedIn’s new AI-detection flag is trained on similar patterns, so if a line trips a human reader, it will likely trip the platform too.

Does AI-generated content actually get less engagement on LinkedIn? On average, yes. A July 2026 study found fully AI-generated LinkedIn posts pulled about 45% less engagement than human-written ones, though AI-assisted posts did better than average in the leadership and inspiration category specifically.

Is it ever fine to use AI for LinkedIn content? Yes, for structuring your thinking and for repurposing a post that already worked. Where it breaks down is letting AI generate the actual sentences and specific claims unsupervised.

What is LinkedIn’s new “AI slop” flag? A feature LinkedIn has confirmed it is rolling out that lets members flag posts and comments that read as AI-generated, part of what the company has called a top priority after outside studies found it carries more fully AI-written content than any other major platform.

How often should a founder repurpose old LinkedIn posts? Aggressively. Only a small share of your audience sees any single post, so reworking a post that already worked, with a new hook and subhook, is smart reach math.