The LinkedIn AI slop button is now live: users who suspect a post was generated by a machine can tap ‘Seems like AI slop’ to flag it, in what the Microsoft-owned professional network is billing as the centrepiece of a broader crackdown on low-quality, automated content.
Hari Srinivasan, LinkedIn’s Chief Product Officer, announced the move on Thursday. ‘AI slop is a top priority for all of us,’ he wrote. ‘People come to LinkedIn to connect with real people and share their real perspectives, ideas, and expertise.’
How the LinkedIn AI Slop Button Works
The feature is not purely cosmetic. According to Yahoo Finance, flags from the button will directly influence how much reach a post receives beyond its author’s own network. In effect, crowd-sourced suspicion becomes a throttle on distribution.
The LinkedIn AI slop button also feeds a separate technical layer. LinkedIn is deploying new classifiers to identify low-quality or AI-generated posts before they surface in suggested content from outside a user’s network, and the flagging data will be used to tune those models over time.
At the account level, LinkedIn will begin privately alerting users when their content is being perceived as inauthentic due to heavy AI use. The company frames this as a corrective nudge rather than a penalty, aimed at people who rely on AI to polish their own drafts rather than those outsourcing their voice entirely.
On the automation side, LinkedIn says it now blocks ‘hundreds of thousands’ of automated comment attempts each day, and millions of other automation attempts over the past couple of months.
Removing the Tool That Made the Problem Worse
Perhaps the most telling structural change sits alongside the button: LinkedIn is pulling its ‘Enhance your post’ feature, which used AI to rewrite users’ text. Mashable describes the tool as having been relatively unpopular before its removal. It will be replaced by a proofreading function that corrects errors without altering a writer’s voice.
Srinivasan offered a candid explanation for why the original feature existed at all. Users post with AI, he wrote, because ‘LinkedIn isn’t a one-word kind of place and they feel more confident running their posts through AI.’ The replacement tool is designed to address that anxiety without stripping out the human element.
LinkedIn is also expanding profile and page verification, and adding an option to block comments from company pages users no longer wish to hear from.
A Wider Industry Scramble for Detection
LinkedIn’s announcement arrives in the same week as a funding round that underlines how commercially urgent the detection problem has become. Pangram, an AI-content identification company, closed a $9 million raise led by Menlo Ventures, with Haystack, ScOp Venture Capital, Script Capital, and Cadenza also participating, according to FinSMEs. The round follows approximately $4 million in seed funding raised in June 2025, shortly after the company launched in 2024.
Pangram, led by CEO Max Spero and CTO Bradley Emi, both Stanford AI and machine learning graduates, claims its Pangram 4 model can identify AI-assisted and mixed human-AI writing with accuracy above 99% and a false-positive rate of 0.0041%. The company competes with Winston AI, Originality.ai, Copyleaks, and GPTZero. Its new capital will go towards improving text detection accuracy and expanding into image analysis.
Substack struck a partnership with Pangram last week to let readers request an AI scan of content on the platform. The integration carries constraints: Axios reports it only operates on text longer than 100 words published after 21 July 2026, when the feature was first announced, and an Android version is still forthcoming. Separately, Substack CEO Chris Best announced that creators will gain access to a ‘How I make this’ statement, allowing them to disclose their AI use to readers directly.
The pressures driving all of this are not slow-moving. Internet infrastructure firm Cloudflare has reported that bot traffic now outpaces human-generated requests on the web, a milestone it says arrived faster than predicted. Digg shut its Reddit competitor in March after concluding it could not contain the volume of bot activity on the site.
For LinkedIn, the harder question is whether user-generated flags and classifier updates can outpace the tools producing the content in the first place. The answer may depend less on the button than on whether the reach penalties attached to it prove sharp enough to change poster behaviour.
