LinkedIn spent two years building AI writing tools into its compose box. On 30 July it began asking members to flag the results. A new option in the three-dot menu on every feed post reads “Seems like AI slop”: one tap thanks the reporter, hides the post from their feed, and feeds the judgement into the classifiers LinkedIn uses to decide what travels.

The change lands on the platform where most B2B marketing teams do their organic publishing, and it formalises something that had been informal until now: machine-written posts carry a distribution cost as well as a reputational one. Hari Srinivasan, LinkedIn’s chief product officer, confirmed the rollout in a post the same day, writing that “AI slop is a top priority for all of us.”

What does the ‘Seems like AI slop’ flag do?

The flag appeared first in LinkedIn’s mobile app, sitting alongside the existing report options on every post. Selecting it produces a short acknowledgement, “Thanks for letting us know. Your feedback helps improve the feed”, and removes the post from the reporter’s feed. 404 Media, which tested the feature on 30 July, reported finding a plausible candidate within roughly six seconds of scrolling.

Member reports do two jobs. They train LinkedIn’s detection models: Srinivasan said slop is hard to define and the definition keeps shifting, so human judgements give the company’s classifiers something concrete to tune against. Those classifiers also shape suggested content, the posts LinkedIn recommends from outside a member’s own network, which is the surface where generated content had the most reach to gain.

The reports will also travel back to the poster. LinkedIn is testing a private flag in creators’ analytics dashboards showing when members felt a post came across as inauthentic or leaned heavily on AI. The company frames this as feedback from real people rather than a verdict from an automated detector, and says the goal is to help writers who refine their own thinking with AI understand when the output stops sounding like them.

Why is LinkedIn retiring its own AI writing tool?

LinkedIn is removing the “enhance your post” feature that redrafted members’ writing with AI, replacing it with a proofreader that corrects spelling and grammar while leaving the writer’s voice intact. Dan Shapero, LinkedIn’s chief executive since April, said in a video announcing the changes that the product will shift from helping members create posts with AI to helping them edit what they wrote themselves, because “there is just no shortcut to figuring out something unique to say.”

The reversal is striking given where the tools came from. LinkedIn is owned by Microsoft, OpenAI’s largest backer, and has spent two years shipping AI writing features, including to Premium subscribers. The line the company now draws runs between AI as an editing aid and AI as a substitute for having a perspective. Shapero’s framing was that the feed’s best content mixes insight and personality, and that insight comes from a person’s own experience rather than from a model.

The new flag reinforces a policy change from May, when LinkedIn said it would reduce distribution for content that appears machine-generated and generic. Flagged posts are not deleted and accounts are not suspended; the penalty operates through reach.

What changes for marketing teams posting on LinkedIn?

The scale figures explain why LinkedIn moved. Detection firm Pangram estimated in early July that 41% of long-form LinkedIn posts and 30% of short-form posts were likely AI-generated, figures first reported by 404 Media, and described LinkedIn as the most AI-saturated of the major social platforms. LinkedIn says its automated defences already block hundreds of thousands of automated comment attempts each day and have stopped billions of automation attempts in recent months, on the company’s own figures.

For marketing teams the practical read is direct. Posts generated wholesale by AI now face member policing, classifier suppression and private inauthenticity warnings, on top of readers who had already learned the tells. Using AI earlier in the process, to pressure-test an argument, tighten a draft or proofread the final copy, is the behaviour LinkedIn’s own product now models. Teams publishing a genuine point of view stand to gain relative visibility as synthetic volume is suppressed, which moves the feed’s economics towards marketers who were writing their own posts all along.

The update also expands access to profile and page verification tools, and adds an option for members to block comments from company pages they no longer want to hear from, a detail worth noting for brand teams that comment widely from their page.

LinkedIn has not said when the analytics flag will move beyond testing, nor whether a member-flagged post loses distribution directly; for now the reports train the models, and the reach penalties introduced in May remain the enforcement.