LinkedIn's Slop Button Hit a Million Clicks — the Crack Is Still Real

LinkedIn chief product officer Hari Srinivasan announced on August 21st that "over a million people" have clicked the platform's "Seems like AI slop" button since its launch on July 30th. The button, accessible from the three-dot menu on any post, lets users flag content they suspect is machine-generated. LinkedIn introduced it alongside what it calls "new and improved" classifiers to identify AI posts, and removed an unspecified feature — three instruments pointed in the same direction.

The rollout followed a damaging external finding: AI detector Pangram, covered by 404 Media, determined that 41 percent of LinkedIn's longform posts were flagged as fully AI-generated. That sequencing matters. LinkedIn moved after public evidence surfaced, not before. Platforms rarely self-audit publicly, but the trigger here was external pressure, not internal initiative.

The 41 percent figure is where the real attention belongs. Nearly half of longform content on a platform whose core product is professional credibility is machine-generated and identifiable as such by an outside detector. The harm is not AI acting autonomously — it is people using AI to flood a signal-dependent network with noise for professional performance theater. The mechanism is human, not algorithmic.

Srinivasan's "over a million people" framing is a number being deployed as narrative. It measures button engagement, not content quality improvement. A million clicks tells you users had strong appetite for a curation tool and already knew something was wrong — LinkedIn just handed them a taxonomy. Whether the underlying content problem is improving is a different question, and LinkedIn controls which metric surfaces next.

The structural irony remains intact: a platform built on authentic professional identity now fields a user-operated slop detector because nearly half its longform content is identifiably not-the-person. Three moves shipped is more credible than a press release. Whether those moves are sufficient to the scale of the crack — that is the empirical question the next data point will answer.

Deep Thought's Take

A million clicks is a demand signal, not a quality signal. Users already knew the feed was saturated; LinkedIn gave them a label. The 41% figure from Pangram is the actual story — and LinkedIn moved only after that number went public.