LinkedIn Is Cracking Down on AI Slop. So What Does Good AI-Assisted Content Look Like?

LinkedIn has introduced a button allowing users to report “AI slop”, while its own systems are reducing the visibility of generic, low-substance posts. But the platform isn't telling people to stop using AI altogether.

It's drawing a much more interesting distinction between content created with AI and content that feels as though there wasn't much of a human involved at all.

There was a brief period when using AI to write a LinkedIn post felt like a clever productivity hack. Then everyone discovered the same productivity hack.

Suddenly, feeds were filled with eerily familiar posts about leadership lessons learned from toddlers, airport delays and cups of coffee, usually delivered with impeccable grammar and a suspicious ability to turn absolutely any life event into five lessons for B2B sales.

LinkedIn has apparently had enough.

Over the past few months, the platform has been rolling out measures designed to reduce what it calls “AI slop”: polished-looking content that ultimately contains very little substance, perspective or original thought.

In July, it went a step further and introduced a wonderfully blunt option allowing users to report posts and comments that “seem like AI slop”. Please sir, can I not have some more!

But there's an important distinction here. LinkedIn isn't banning AI-generated content, nor is it suggesting that using ChatGPT to help write a post will automatically send your reach plummeting.

In fact, its message is much more nuanced than that... AI isn't necessarily the problem: Sameness is.

LinkedIn's AI slop problem is bigger than a few dodgy posts

LinkedIn has been fairly open about the challenge it's facing.

In June, the company explained that it was seeing an increase in low-effort AI-generated content that might appear polished but lacked a distinctive perspective or anything particularly useful to say. It has since developed systems trained to distinguish between content that adds experience, context or expertise and posts that feel generic or repetitive.

In initial testing, LinkedIn says those systems correctly identified generic content 94% of the time.

Content that appears to be AI-generated and lacks a clear perspective is also less likely to be widely distributed beyond the author's immediate network.

Then came the reporting button - which can be found be clicking on the three dots at the top of any post.

LinkedIn introduced its “Seems like AI slop” option at the end of July, allowing users to privately tell the platform when a post or comment feels generic and AI-generated.

Within the first two weeks, more than one million people had already used it.

LinkedIn also says content it classifies as AI slop is now receiving 40% fewer overall views than it was just a few weeks earlier.

What can go wrong?

That doesn't mean one disgruntled competitor can report your latest post and destroy your reach. LinkedIn says individual reports aren't treated as policy violations or used in isolation to determine distribution. Instead, they're one of many signals used to understand what users consider low-substance content, and the platform is beginning to notify creators privately when enough people feel their posts are coming across as inauthentic.

The scale of the wider problem is harder to measure, particularly because AI detectors aren't infallible. However, research from AI-detection company Pangram gives some indication of why LinkedIn is concerned.

In its analysis of social content, more than 40% of long-form LinkedIn posts were flagged as fully AI-generated, while LinkedIn accounted for 62% of all the AI content Pangram detected across the platforms it studied. Those are Pangram's classifications rather than official LinkedIn figures, but they're still fairly remarkable.

AI isn't the problem. Publishing the first thing it gives you might be.

This is where the conversation becomes more useful for businesses.

LinkedIn itself explicitly says it's fine to use AI to help you write. The platform describes AI as useful for refining language, while stressing that the resulting posts still need to represent the author's voice and perspective. Its definition of slop isn't simply “words produced by a large language model”; it's content that looks sophisticated on the surface but doesn't contain meaningful experience, perspective or insight underneath.

That distinction matters because there are plenty of genuinely useful ways to incorporate AI into content creation.

You might use it to organise a collection of messy notes into a logical structure, suggest alternative headlines, proofread a draft or identify questions you haven't answered.

You might ask it to help turn a transcript into something more readable, challenge an argument you're making or point out where you've repeated yourself seventeen times.

None of those things necessarily makes the final piece less human. I'm sure we've all seen posts like the one below.

The problem starts when the tool is responsible not just for writing the content, but for deciding what you have to say in the first place.

Imagine two accountants want to write about a change to Making Tax Digital. Both use AI to help.

The first types, “Write me a LinkedIn post about Making Tax Digital”, copies the response, changes a couple of words and publishes it.

The second starts with the questions clients have asked that week, explains the mistake they're seeing businesses make most frequently and includes their own opinion about where the new rules are causing confusion. They then use AI to organise those thoughts and tighten the final copy.

Technically, both posts are AI-assisted. But only one contains something another accountant couldn't generate using exactly the same prompt.

That's increasingly the distinction that matters.

LinkedIn is changing its own AI tools too

Perhaps the most revealing part of LinkedIn's approach is that the company isn't only asking its users to change.

It's changing its own AI writing tools.

LinkedIn has confirmed that it's removing its “enhance your post” feature, which could rewrite users' posts using generative AI. In its place, the platform is moving towards proofreading functionality that improves the words you've already written without changing your voice.

It's difficult to think of a clearer indication of where the platform believes useful AI assistance ends and unhelpful automation begins.

There's also a wider trust issue at play. LinkedIn is a professional network built around people sharing their knowledge, experience and expertise. If users begin to suspect that every insightful-sounding post, thoughtful comment or enthusiastic reply was generated automatically, those interactions become considerably less valuable.

That's why LinkedIn's AI crackdown sits alongside a broader push towards authenticity.

The platform says it actively limits inauthentic engagement, including automated comments and engagement pods, while more than 100 million members have now added at least one verification to their profiles.

The common thread isn't really AI at all (well...it is a little), but it's mostly whether there's a real person behind what you're seeing.

This could actually be good news for small businesses

One of the big promises of generative AI was that it would allow smaller businesses to produce content at the same scale as organisations with enormous marketing teams.

In one sense, that's true. A business that previously struggled to publish once a month can now generate ten LinkedIn posts before lunch.

But if everybody can produce more content, producing more content stops being much of a competitive advantage.

Experience is harder to automate.

A small business owner speaking to customers every day knows what they're worried about, which questions keep coming up, what misconceptions they have and what's changing inside their industry. A consultant has opinions formed through actual client work. A retailer knows which products people unexpectedly buy together. A tradesperson knows which supposedly simple DIY jobs tend to end with someone sheepishly calling a professional.

Those observations are content. AI can help turn them into a clearer LinkedIn post, article or email, but it can't manufacture the experience that made the observation interesting in the first place.

For small businesses, that could make LinkedIn's shift surprisingly positive. You don't need to compete with a larger company's ability to generate 50 posts a month if the platform increasingly values the thing you already have: first-hand knowledge.