Picture a marketing director who has posted on LinkedIn five times a week all year, using an AI tool to turn a one-line idea into a polished post each morning. "7 lessons I learned from failing." "The one mindset shift that changed my career." For months, the posts did fine. Since the summer, impressions have fallen by more than half. Then a small notice appears in her analytics: some readers feel her content comes across as inauthentic. The week after, she posts something different: a rough, specific account of a campaign that went wrong and the awkward conversation with her CEO that followed. It becomes her best post of the year. (This is an illustrative scenario, not a documented LinkedIn case.)
LinkedIn has spent 2026 turning against what everyone now calls AI slop: posts that sound polished, say nothing, and could have been written by anyone about anything. It's not removing that content, but it is limiting how widely it's distributed. The shift is easy to misread as "LinkedIn penalises AI". It doesn't. LinkedIn isn't punishing AI; it's punishing posts anyone could have written. The test that matters now isn't who, or what, wrote the post. It's whether only you could have written it.
This post covers LinkedIn's changes and what they mean for organic reach. Broader strategy for the platform is covered in the state of LinkedIn in 2026.
What LinkedIn changed
The crackdown came in three steps. On 4 June 2026, LinkedIn's newsroom published "Keeping conversations real on LinkedIn", first posted on LinkedIn on 20 May, targeting low-quality AI-generated posts, generic AI comments and automation tools that produce content at scale. Content that appears AI-generated and lacks a clear perspective would not be removed, but would be less likely to be widely distributed beyond the author's immediate network.
On 30 July, LinkedIn announced a "Seems like AI slop" feedback option for posts and comments, which rolled out in August. The company's chief product officer described AI slop as a top priority. LinkedIn said member feedback would help inform its content-quality classifiers, that authors would be told privately, in their analytics, when people perceive their content as inauthentic because of heavy AI use, and that it would replace its own AI "enhance your post" feature.
On 15 September, LinkedIn reported that its recent efforts had helped reduce views of content it classifies as AI slop by 40%. That's LinkedIn's own figure, and it doesn't show the feedback option alone caused the drop. Separately, more than a million members used the option in its first two weeks. LinkedIn also described replacing "enhance your post" with tools that proofread and improve clarity without changing a member's voice, a useful signal of where it draws the line.
SocialBee
Recycles your best content automatically keeping your social media consistently active
Best for: Content Recycling Tool
How LinkedIn defines slop
LinkedIn's guidance on AI-assisted content describes AI slop as low-effort, likely AI-generated material that may look polished but lacks a clear point of view, a distinctive perspective or real substance. It also points to content that's generic, repetitive, recycled or designed mainly to game attention. Notice what's missing from that definition: any ban on AI assistance. The problem is emptiness, not the tool.
The scale of the problem explains the urgency. An Originality.ai study of 3,368 long-form posts from 99 influential profiles classified 53.7% of 2025 posts as "Likely AI" using its own detector and threshold. That's a selected sample, not every LinkedIn post, and "Likely AI" is a detector classification, not proof a post was entirely AI-written. When so many posts share the same structure, the same hooks and the same safe insights, readers stop reading, which is exactly what a feed can't afford. The same problem shows up everywhere AI writes by default, as covered in the content sameness problem.
How the penalty works
The enforcement is limited distribution, not deletion. Your connections can usually still see the post, but LinkedIn's classifiers can limit how widely it's recommended beyond your network. For most accounts, that's where growth comes from.
A post normally reaches 10,000 people:
→ 3,000 from your own network
→ 7,000 through recommendations to people outside it
In the extreme case, if the post stopped being recommended beyond your network entirely, reach would fall towards 3,000, a drop of around 70%. This is a hypothetical scenario, not a measured or guaranteed LinkedIn penalty.
The hidden cost: those 7,000 out-of-network readers are the people who don't know you yet, which is where new followers, customers and opportunities come from. Limited distribution doesn't just lose reach; it can slow your growth.
The split varies widely by account and post. Check your own analytics for how much reach comes from outside your network.
Member feedback isn't an automatic penalty. It helps inform LinkedIn's content-quality classifiers, and a post that receives enough community feedback may trigger a private tip in the author's analytics. But LinkedIn's chief product officer has said that no single piece of feedback determines distribution: LinkedIn weighs many signals together, with safeguards against misuse. Comments are in scope too. Auto-generated replies that simply restate the post they're under, or say "Great insights, thanks for sharing", are the pattern LinkedIn has said it's targeting.
The test LinkedIn isn't punishing AI; it's punishing posts anyone could have written. The question that matters now isn't who or what wrote the post. It's whether only you could have written it.
What gets throttled, and what still earns reach
| Likely throttled | Still earns reach |
|---|---|
| Generic lessons and listicles anyone could write | A specific story from your own work, with real details |
| Safe, agreeable "insights" | A clear point of view, including disagreement |
| Recycled trends and summaries | Your own data, screenshots or observations |
| Auto-generated comments | Comments that add context or experience |
| High-volume posting from short prompts | Fewer, better posts built on real material |
| Engagement bait and pods | Questions you genuinely want answered |
First-hand experience. The strongest posts describe something that happened to you: a decision, a mistake, a result, a conversation. Real details, such as numbers, constraints and what you'd do differently, are what make a post impossible to fake.
A clear point of view. Saying something people might disagree with, or explaining how you changed your mind, gives readers a reason to stop. Generic AI writing tends to avoid positions; people don't.
Original material. Your own data, customer observations, before-and-after examples and screenshots give readers something they can't get elsewhere. Sales conversations are a rich source, as covered in building a content engine around sales objections.
Real people. Posts from founders and employees often travel further than brand announcements, a tendency rather than a confirmed ranking rule, and buyers increasingly trust people over brands, as covered in why buying committees trust peers over ads. The approach for leaders is in building a founder-led LinkedIn presence, and for wider teams in employee advocacy for B2B.
Thoughtful comments. A comment that adds a relevant example or a respectful counterpoint builds visibility and relationships at the same time. Write fewer, better ones, in your own words.
How to use AI without producing slop
LinkedIn's guidance says AI can help with refining language, proofreading or being concise, but content should reflect your voice, perspective and experience, and you should review anything AI helped with. It doesn't ban AI-written posts; the risk it describes is generic, low-substance content and over-reliance on automation, most likely when a post starts from nothing. A practical workflow looks like this. Start with a real idea from your work, and note the specific details that make it yours. Write a rough draft in your own words, even if it's messy. Use AI, if you like, to tighten wording, check clarity or suggest a stronger opening. Then read the final version aloud and ask: could anyone else have written this? If the answer is yes, it needs more of you in it.
For teams, the same principle belongs in your process. Templates that require real inputs, such as customer quotes, numbers and first-hand examples, produce far better drafts than open-ended prompts, as shown in prompt templates for marketers. Keep a human editor, keep your voice consistent using an AI content workflow that keeps your brand voice, and set clear rules on AI use and disclosure, as covered in AI governance for marketing teams. Remove any automation that comments or posts on anyone's behalf without review.
The same principle is spreading well beyond LinkedIn. Inboxes are filling with AI-written outreach, and the emails that still get answered are the ones that could only have been sent to one person, as covered in cold email that gets replies. Across B2B marketing, AI is making average content cheap and abundant, which makes genuinely original perspective more valuable, a shift explored in how AI is changing demand generation. LinkedIn is simply the first major platform to build that judgement directly into its feed.
What to do this month
Check your analytics for any notice about inauthentic content, and look at how your reach splits between your network and beyond it. Review your last twenty posts and mark those that anyone could have written; those are your slop candidates. Replace volume with substance: as a practical rule, two posts a week built on real material are a better bet than five generic ones. Turn off auto-commenting tools. And give your team permission to post things that are specific, a little rough and genuinely theirs. Brands that want their LinkedIn presence rebuilt around authentic voices often work with a specialist social media marketing team.
The short version
Through 2026, LinkedIn has moved against AI slop: in June it published measures against low-quality AI posts, generic AI comments and automation tools; in July it announced a "Seems like AI slop" feedback option, rolled out in August, plus private notices to authors and the replacement of its AI post-enhancement feature; and in September it said its efforts had helped cut views of content it classifies as slop by 40%, while more than a million members used the option in two weeks. Slop isn't deleted, but its distribution beyond your network can be limited, which can cut reach and slow growth; feedback is one signal among several, not an automatic penalty. LinkedIn defines slop as polished content without a point of view, perspective or substance. LinkedIn isn't punishing AI; it's punishing posts anyone could have written. What still earns reach is first-hand experience with real details, a clear point of view, original material, posts from real people and thoughtful comments. Use AI to refine your own draft, not to generate posts from nothing, remove auto-commenting tools, and ask of every post: could only I have written this?
Has your LinkedIn reach dropped this year?
We help brands and leaders build LinkedIn content that people stop for.
Explore Social Media Marketing →Frequently asked questions
Is LinkedIn penalising AI-generated content?
LinkedIn is limiting the reach of low-effort, generic AI content rather than penalising the use of AI as such. In June 2026 it published measures against low-quality AI-generated posts, generic AI comments and automation tools, and in July it announced a 'Seems like AI slop' feedback option, rolled out in August, alongside classifiers that help identify low-quality content. Content identified this way is not removed; instead, it is less likely to be widely distributed beyond the author's own network. LinkedIn describes AI slop as material that may look polished but lacks a clear point of view, distinctive perspective or substance, or that is generic, repetitive, recycled or designed mainly to game attention. Posts that use AI to help refine a genuine idea, with the author's own experience and perspective, are not the target.
What is the 'Seems like AI slop' button on LinkedIn?
It is a feedback option LinkedIn announced in July 2026 and rolled out in August that lets members say a post or comment feels like low-value, AI-generated content. LinkedIn says this feedback helps inform its content-quality classifiers, and its chief product officer has said no single piece of feedback determines how content is distributed. If a post receives enough community feedback, its author may see a private tip in their post analytics. According to LinkedIn, more than one million members used the option within the first two weeks. Separately, in September 2026 LinkedIn said its efforts had helped reduce views of content it classifies as AI slop by 40%.
What kind of LinkedIn posts still get good reach?
Posts that only the author could have written. That typically means content rooted in first-hand experience, such as a specific project, decision, mistake or result, with real details rather than general advice. A clear point of view helps, including disagreement with common wisdom or an honest account of changing your mind. Original material, such as your own data, screenshots, examples or observations from customers, gives readers something they cannot get elsewhere. Thoughtful comments that add context to other people's posts also build visibility and relationships. Posts from real people, such as founders and employees, tend to outperform generic brand announcements. None of this rules out using AI for research or editing, but the ideas, experience and judgement need to be genuinely the author's own.
Can you still use AI to write LinkedIn posts?
Yes, as a helper rather than a replacement for your own thinking. LinkedIn's guidance says AI can help refine language, proofread or make content more concise, as long as it reflects your own voice, perspective and experience and you review it before posting. LinkedIn hasn't banned AI-written posts; the risk it describes is generic, low-substance content and over-reliance on automation. The risk comes from generating whole posts from a short prompt with no personal input, publishing at high volume, or using tools that automatically comment on other people's posts. Start with your own idea, experience and opinion, write a rough draft in your own words, use AI to help edit if useful, and read the final version carefully to make sure it still sounds like you and says something specific.
