LinkedIn Slop Detector's One Millionth Flag Was a Human Who Has Typed Her Own Posts Since 2009

SUNNYVALE—LinkedIn's "Seems Like AI Slop" button reached one million user clicks Monday, which the company described as a milestone for platform authenticity. Among the accounts suspended was Sandra Kellner, a 44-year-old supply chain director from Naperville, Illinois, who has posted on LinkedIn without AI assistance since 2009.

Kellner's suspension notice informed her that 94% of her historical posts exhibited "patterns consistent with synthetic thought leadership content." Phrases flagged in her account's risk profile included "excited to share," "proud of the team," "here for it," and "lessons learned" — all of which Kellner confirms she typed herself, between 2011 and last Tuesday.

LinkedIn's detection model, trained on 847 million posts previously flagged as AI-generated, assigned Kellner a 94.7% probability of being non-human, placing her in the platform's top decile for suspected synthetic accounts.

Kellner appealed. The appeal was reviewed by a separate AI. It was denied in four minutes.

LinkedIn's false-positive rate is "under 12%," a spokesperson confirmed, which the company projects will incorrectly suspend approximately 120,000 authentic human users per million flags. Kellner was offered a video verification process in which a third AI analyzes recordings for signs of scripted delivery.

Her video was denied in three minutes.

"We take the authenticity of our community seriously," the spokesperson said. "Sandra's content spoke for itself."

Kellner's account remains suspended. The spokesperson's statement has not been flagged.