In September, LinkedIn did something unusual for a platform that sells AI writing tools: it published a newsroom post titled “How LinkedIn is tackling AI slop”. A month earlier it had added a button letting members flag posts that “seem like AI slop”. Perhaps alarmingly, more than a million people pressed it in the first few weeks.
When a platform starts building buttons to report its own content, it’s fair to say there’s a bit of a problem. It’s also a striking turn in the LinkedIn AI narrative, which until recently was all about helping members post more. The more useful question is what it means for brands trying to be heard there and for executives hoping to use the platform as an effective thought leadership channel.
Key takeaways
-
LinkedIn defines AI slop as “low-effort, likely AI-generated content that may sound polished on the surface”, and has introduced a reporting button and classifiers to reduce its reach.
-
Estimates of AI-generated long-form posts on LinkedIn range from around 40% to over 80%, though detection methods vary and none are official.
-
LinkedIn contributed to the problem through built-in AI writing tools, gamified collaborative articles and engagement-driven distribution.
-
Audiences rated most thought leadership poorly long before ChatGPT; AI has multiplied the volume of generic content.
-
For brands, the risk is reputational for executives, reach-related for company pages and scale-related for employee advocacy. AI used to sharpen a genuine view is fine; AI used to replace one is what audiences and LinkedIn are filtering out.
What LinkedIn means by AI slop
LinkedIn’s own definition is “low-effort, likely AI-generated content that may sound polished on the surface”. On LinkedIn, it’s pretty hard to miss… the post that opens with a one-line hook, tells a tidy anecdote about a taxi driver, lists five lessons and ends with “Agree?”. Grammatically perfect, broadly sensible, littered with emojis and devoid of real insight or believability.
The wider culture has caught up too. Merriam-Webster named “slop” its word of the year for 2025, and Australia’s Macquarie Dictionary chose “AI slop” outright. LinkedIn’s chief product officer, Hari Srinivasan, has called it “a top priority for all of us”.
How big is the problem?
Nobody outside LinkedIn knows exactly, and LinkedIn hasn’t published a figure. The estimates that exist come from AI-detection companies, so treat them as indications rather than facts, as these companies obviously have their own motivations for perpetrating a narrative that AI content is ubiquitous!
Originality.ai estimated in late 2024 that 54% of long-form English posts were likely AI-generated, and in July 2026 put the figure at 81% of a 5,000-post sample. Pangram Labs, analysing more than a million items across platforms this summer, found LinkedIn produced 62% of all the AI content it flagged, despite being only about a third of what it scanned.
The methods vary and detectors make mistakes, particularly with non-native English writers, but every study points the same way, and anyone who scrolls the feed can feel it.
LinkedIn helped create this
It would be a little rich for LinkedIn to take all the credit for cleaning this up. Since 2023 it has given Premium members tools to draft and “rewrite” posts with AI, built directly into the posting box. The same year it launched collaborative articles, which used AI-generated prompts and rewarded contributors with a gold “Community Top Voice” badge. That produced more than 10 million contributions in a year, and so much filler that LinkedIn retired the gold badge in late 2024, admitting it was “challenging to maintain the highest quality standards”.
Underneath it all sat a feed that rewarded, and continues to reward, volume and engagement, which encouraged engagement pods, automated comments and the formulaic posts that perform well by those measures. This year LinkedIn has been unwinding all of it: cutting reach for engagement bait and “recycled” posts, shutting down pods and comment automation, replacing its “enhance your post” tool with a proofreader that keeps “a member’s voice”, and reporting a 40% drop in views of content it classifies as slop.
The quality problem is older than ChatGPT
The part that often gets glazed over is that audiences were increasingly unimpressed with a lot of thought leadership content long before generative AI arrived. In the Edelman and LinkedIn B2B Thought Leadership study published in 2021, a year before ChatGPT launched, only 15% of decision-makers rated the thought leadership they read as very good or excellent, and 71% said less than half of it gave them any valuable insight. The 2020 edition found 28% rating it mediocre to very poor.
AI didn’t create bland content, far from it. What is has enabled, however, is the swifter production of bland content far faster and cheaper than before; multiplying the volume of something audiences were already tired of. This perhaps explains why the backlash has been so quick.
What it means for brands
Anyone planning LinkedIn for brands is really managing three channels at once, and the risk looks different in each.
For executive profiles, the risk is reputational as well as algorithmic. LinkedIn’s chief executive, Ryan Roslansky, has pointed out that the bar is higher there than elsewhere because “this is your resume online”. Research backs the instinct. A 2025 study in Organizational Behavior and Human Decision Processes, covering 13 experiments with more than 5,000 participants, found people who disclose AI use are trusted less, and that being found out by someone else is worse than admitting it. Ghostwriting is nothing new, and earlier research found audiences accept it when it’s open. What they don’t forgive is a voice that turns out not to be the executive’s at all.
For company pages, the challenge is reach. Organic company page posts already reach a small share of followers, and LinkedIn now lets members block comments from company pages they don’t want to see. A brand page posting generic AI-generated content is spending its scarce attention on posts nobody will remember.
For employee advocacy, the risk is scale. If a programme hands staff the same AI-drafted post to share, it now looks exactly like what LinkedIn is building systems to detect and downrank. The value of employee advocacy has always been that real people say things in their own words. Templated content throws that away.
The case for AI
None of this is an argument against using AI. LinkedIn itself is careful to say that “AI and slop are not the same thing”, and its own guidance is that “it’s ok to use AI to help you write, but your posts and comments need to represent your voice and your perspectives”. A widely quoted Massachusetts Institute for Technology (MIT) study found professionals using AI completed writing tasks 40% faster with quality judged 18% higher, with the biggest gains for weaker writers.
Used for research, structure, editing, translation and accessibility, AI makes good LinkedIn thought leadership easier to produce; the problem starts when it supplies the opinion as well as the final touches or polish.
The brands that stand out in a feed full of AI generated content will be the ones with something specific to say: a real view, earned through experience, said in a recognisably human voice. That has always been the job of thought leadership. It has simply become much easier to spot who isn’t doing it.
FAQs
About the author
Simarin Tandon | Junior Digital Account Director
Having worked with brands across the Beauty & Wellness, FMCG, FinTech, and Home & Lifestyle sectors, Simarin focuses on driving acquisition and growth, whilst managing the Digital team at brandnation.
A curious marketer, Simarin’s finger is always on the pulse when it comes to performance and digital updates across both paid and organic platforms.
LinkedIn describes AI slop as “low-effort, likely AI-generated content that may sound polished on the surface”, including posts that feel generic, repetitive or recycled.
LinkedIn says it uses classifiers to limit the distribution of content it identifies as AI slop, and reported a 40% reduction in views of that content in 2026. It says using AI to help write is acceptable as long as posts reflect the member’s own voice and perspective.
Ghostwriting is widely used and generally accepted when it reflects the executive’s genuine views. Research suggests trust drops when audiences discover content wasn’t really the executive’s, so the safest approach is a ghostwriter who captures the leader’s actual opinions and voice.
There’s no official figure. Third-party detection studies have estimated between 40% and 81% of long-form posts, though these methods vary and can misclassify human writing.
Use it for research, structure, editing, translation and accessibility, and keep the opinions, experience and voice human. Avoid templated, AI-drafted posts for employee advocacy, which now resemble the content LinkedIn is downranking.
