LinkedIn says its “Seems like AI slop” feedback option has been selected more than 1 million times since it launched on July 30, an early sign of how actively members are responding to generic or low-substance content in the professional network’s feed. Chief Product Officer Hari Srinivasan disclosed the figure in an August update, saying the selections are helping LinkedIn understand how its community experiences low-quality content. The number should not be read as a count of 1 million unique people, distinct posts, or confirmed instances of AI authorship. (LinkedIn)
The option appears in the three-dot menu on posts. LinkedIn has characterized it as a feedback and disinterest signal rather than a formal route for reporting policy violations. Sam Corrao Clannon, the company’s creator product lead, said LinkedIn defines “AI slop” as content that may look polished but lacks experience, perspective, or insight—an intentionally different standard from simply determining whether software assisted in the writing. (LinkedIn)
That distinction matters because the button’s label can imply a provenance judgment that the system is not designed to make. LinkedIn says people may use AI to refine language or improve expression without running afoul of the effort. Its stated focus is whether a post feels generic, empty, or low-effort to readers, not whether every sentence was drafted without generative AI. In practice, that places the new control closer to a quality signal than an authorship verdict. (LinkedIn)
How LinkedIn says the feedback signal works
LinkedIn also says a single selection does not determine distribution for a post. Srinivasan said the company combines many signals and has safeguards intended to prevent individual feedback from unfairly targeting members. LinkedIn has also said the rollout uses member feedback alongside classifiers intended to identify AI slop or generally low-quality content. The company has not publicly detailed the weighting of those signals, the thresholds that trigger creator notices, or the accuracy and error rates of its classifiers. (LinkedIn)
The company is beginning to show some authors a message in Post Analytics when a post receives enough community feedback, Srinivasan said. LinkedIn also says members are now seeing 40% fewer views of content that it classifies as AI slop than they did a few weeks earlier. That is a company-reported aggregate outcome, however, not evidence that the feedback button by itself caused the reduction in views. (LinkedIn)
Context for the rollout
The rollout follows a July analysis by AI-detection company Pangram that found more than 40% of longform LinkedIn posts in its dataset were flagged as fully AI-generated. Pangram’s dataset contained 1,002,627 posts scanned through an opt-in browser extension across several platforms; its measurement reflects the posts its participating users encountered and scanned, not all material published on LinkedIn. The finding is useful context for why platforms are seeking new quality controls, but it is a detector-based estimate rather than a platform-wide audit or proof of the origin of any individual post. (Pangram)
The Verge’s reporting on LinkedIn’s rollout described the initiative as combining member feedback with new or improved classifiers intended to identify AI slop or generally low-quality material. That combined approach reflects the central moderation challenge: human feedback can capture whether content feels useful and authentic, while automated systems can operate at feed scale. Neither signal is definitive on its own, particularly when AI can be used for editing, translation, accessibility support, or drafting that is substantially revised by a person.
What it means for LinkedIn publishers
For professionals who publish on LinkedIn, the immediate message is not that any use of AI is disfavored. LinkedIn’s stated standard is whether a post contributes recognizable expertise, experience, or a point of view. For the platform, the million-selection milestone creates both a valuable stream of feedback and a difficult governance test: translating crowd judgments and classifier outputs into ranking decisions without treating polished writing, unconventional style, or legitimate AI assistance as proof of low-quality thought. (LinkedIn)
