On July 30, 2026, LinkedIn added a button labeled "Seems like AI slop" to every post on the platform. Within five weeks, over one million people had clicked it. LinkedIn's chief product officer Hari Srinivasan confirmed the number on August 28 and said the platform is now showing 40 percent less reach to posts its classifiers flag as AI-generated.

That is not a niche feature used by power users. One million people went out of their way to flag content as machine-generated garbage. The audience is paying attention, and they are tired of it.

The numbers are worse than you think

Pangram, an AI detection company, ran a study using data from its Chrome extension and found that 41 percent of LinkedIn's longform posts — anything over 250 words — were flagged as fully AI-generated. Not AI-assisted. Not lightly edited. Fully machine-written.

LinkedIn was the worst offender, but it was not alone. Nearly half of X/Twitter's longform articles contained AI writing. Even Substack, which prides itself on substantive human writing, saw 21.9 percent of posts flagged as AI-generated or AI-assisted. Across all social platforms, one in four longform posts was fully AI-written.

The content is not subtle. You have seen it. The five-bullet opener. The "In today's rapidly evolving landscape" hook. The perfectly structured paragraphs that say absolutely nothing. The em-dashes. Oh, the em-dashes.

Microsoft's own executive is sick of it

Here is where it gets interesting. Ryan Roslansky, the Microsoft executive who oversees both Office and LinkedIn, posted on LinkedIn last week about receiving a document from a colleague that was clearly AI-generated. His response? He had his own AI summarize it. The result: nothing of value. He called it a "doom loop."

Think about that. The person running the division that ships Copilot into Word, Excel, and PowerPoint — the tools most responsible for flooding workplaces with AI-generated text — is publicly complaining about AI slop. The call is coming from inside the house.

Roslansky wrote: "If we're not careful, we'll end up with a very expensive way to avoid writing and reading in an ocean of sameness, genericness." He is right, and the fact that he is saying it publicly means the problem has crossed a threshold inside Microsoft itself.

Why this matters if you run a business

If you are using AI to draft client proposals, write blog posts, respond to RFPs, or generate marketing copy, you need to understand what is happening. Your clients are developing AI slop radar. They are surrounded by it on LinkedIn, in their inboxes, in the proposals they receive from your competitors. They are learning to spot the patterns, and when they do, trust erodes.

This is not an abstract branding problem. It is a revenue problem. A client who realizes your proposal was ChatGPT output starts wondering what else was automated. Did you actually assess their needs, or did you paste their brief into a prompt? Did you write that strategic recommendation, or did an LLM generate something that sounds strategic but is actually generic?

The Pangram data makes the stakes clear: LinkedIn posts made up a third of all scanned items, yet accounted for two-thirds of all AI-flagged content. The platform where professionals build their reputations is the one most saturated with machine-written text. If your business development depends on LinkedIn content, your posts are competing with an ocean of slop. Standing out now requires being visibly, unmistakably human.

The tools are not the problem

None of this means you should stop using AI. That would be like refusing to use spell-check because some people rely on it too heavily. The tool is not the issue. The issue is output that reads like it came from a tool.

The difference between good AI use and slop is what happens after the draft. Did you add your own experience, your specific client context, your actual opinion? Or did you read the first draft, think "looks good," and hit send?

Roslansky's doom loop is real. Someone writes a prompt, generates a document, sends it to a colleague. The colleague's AI summarizes it. The summary contains nothing new. Everyone spent compute cycles to arrive at nothing. The document existed not to communicate but to check a box.

If your client proposals read like that, you have a problem. If your blog posts read like that, you have a bigger one, because they are public and permanent.

What to do about it

The fix is not complicated, but it requires discipline. Use AI to generate a first draft, then rewrite the parts that matter. Add a specific detail from your last client call. Include a number that is not in any training data. Take a position that a language model would never take because it is too opinionated or too specific.

LinkedIn's algorithm change is the leading indicator. The platform is actively deprioritizing AI-generated content. Google has been doing the same with search rankings for over a year. The platforms are telling you what the audience wants: real thinking from real people.

The million-click slop button is not a feature announcement. It is a market signal. Your audience can tell. They are voting with their clicks. The question is whether your content would survive the same scrutiny.

Sources: The Verge — Over 1 million people have clicked LinkedIn's AI slop button · Pangram — AI Content Is Everywhere on Social Media, Especially LinkedIn · Ryan Roslansky on LinkedIn — AI slop in the day to day of work