I’ll admit it, I’ve scrolled past posts on LinkedIn that started with “In today’s fast-paced digital landscape…” and immediately known, before reading another word, exactly what kind of post I was about to skim past. Apparently I’m not alone, and LinkedIn itself just admitted it has a problem on its hands.
What LinkedIn actually rolled out
LinkedIn quietly added a new option to the three-dot menu on every post: “Seems like AI slop.” Click it, and the post gets hidden from your feed while LinkedIn logs the report to help tune its own detection models. The company’s chief product officer framed it plainly, acknowledging that AI slop is a real, prioritized problem and that user reports help the platform figure out where to draw a line that keeps shifting. In the same announcement, LinkedIn confirmed it’s also removing its own “enhance your post” AI rewriting tool, replacing it with a proofreading feature that’s meant to polish grammar without rewriting someone’s actual voice.
Why this happened now
The timing isn’t a coincidence. Independent AI-detection research using a tool called Pangram had estimated that a large share of long-form LinkedIn posts, and a meaningful chunk of shorter ones too, showed signs of being substantially AI-generated. That kind of public estimate, combined with a competing platform publicly needling LinkedIn over the same issue just days earlier, put real pressure on the company to be seen doing something visible about it, not just quietly adjusting a ranking algorithm behind the scenes.
What actually gets a post flagged
LinkedIn hasn’t published a specific checklist, and its own product lead has said the definition of “slop” shifts over time as detection improves. Based on what’s been reported and what tends to trip up AI detectors generally, a few patterns show up again and again:
- Uniform sentence rhythm. Sentences that all land at a similar length and pace read as mechanically smooth rather than naturally varied, which is one of the most consistent tells AI detection tools rely on.
- Overused transition phrases. Words like “delve,” “leverage,” “crucial,” and “in today’s landscape” have become so associated with AI writing that their presence alone raises suspicion, fairly or not.
- Generic, low-specificity claims. A post that could have been written about any industry, any company, any person, without a single concrete detail, reads as templated even if a human wrote it.
- Structure over substance. A tidy three-point framework with a bolded takeaway at the end is efficient, but it’s also exactly the shape AI tends to default to, and readers have started noticing the pattern.
How to write posts that don’t get swept up in this
The good news is that most of what makes a post read as human is also just what makes it good, worth remembering if the goal is engagement, not just avoiding a flag.
- Lead with something specific to you. A real number, a real mistake, a real moment from your week does more to signal authorship than any amount of careful phrasing.
- Let sentence length vary naturally. Write the way you’d actually explain something to a colleague, some sentences short, some longer, rather than smoothing everything into the same cadence.
- Cut the stock transition words. If a sentence would sound exactly the same without “leverage” or “delve into,” it will read more naturally without them.
- Say the specific thing, not the general category of thing. “Our Q3 signups dropped 12% after we changed the pricing page” reads as real. “We faced some challenges with our conversion funnel this quarter” reads as filler, whether or not AI wrote it.
The bigger shift this is part of
LinkedIn isn’t acting alone here. Newsletter platform Substack added its own AI-detection partnership with Pangram just days before LinkedIn’s announcement, explicitly framing it as not wanting to become “another LinkedIn” full of generic content. Taken together, this looks like the start of a broader shift across content platforms toward actively surfacing and penalizing low-effort AI content, rather than just quietly deprioritizing it in a feed algorithm. For anyone writing professionally online, that’s worth taking seriously well beyond just LinkedIn specifically. Coincidentally, this rollout landed the same week as a very different kind of technology story, our coverage of the water utility cyberattacks making clear that trust and authenticity concerns are showing up across very different corners of the tech world right now.
FAQ
Does the “seems like AI slop” button actually remove the post, or just hide it for me?
It hides the post from your own feed and logs the report for LinkedIn’s internal review, it doesn’t delete the post from the platform or notify the poster publicly, though the poster can see the report privately in their own analytics.
Will using an AI tool to help draft a post automatically get it flagged?
Not necessarily. The concern is with posts that read as generic or templated regardless of how they were produced, a heavily-edited AI draft that includes real specifics and a natural voice is a different thing than an unedited AI output.
Is LinkedIn removing all AI writing help from the platform?
No. It’s replacing its “enhance your post” rewriting tool with a proofreading feature aimed at fixing grammar and clarity without altering the substance or voice of what someone wrote.
Are other platforms doing anything similar?
Yes. Substack recently added AI-detection scanning through a partnership with Pangram, and it’s a reasonable bet more content platforms will introduce comparable tools as pressure around AI-generated content grows.
