A joke bot mass-produces LinkedIn thought leadership, and the engagement arrives anyway

Peter Epstein, who runs a Los Angeles PR firm serving advertising-industry clients, built a website called Cringebot 3000 that deliberately mass-produces grating AI-generated posts for LinkedIn, and it spread across Reddit and Hacker News this week ^[1][2][3]. Business Insider posted the output to LinkedIn and reported, in the words of its own headline, that "Darkness (and engagement) stared back" ^[4].

The interesting part is the specification. The user supplies a topic, which can be a single word, then picks one of eight styles, among them "The Vulnerability Post (oversharing with a motivational twist)", "The Inspiring Story (finding life lessons everywhere)" and "The Corporate Dropout (quit the 9-5 to follow my passion)" ^[5][6]. Business Insider typed "AI in journalism" and selected "You Go Girl! (empowerment culture and smashing glass ceilings)" ^[7]. What came back was a meandering story with a grippy opening sentence and a corny kernel of truth at the end, loaded with hashtags and carrying a disclaimer at the bottom stating it was written by Cringebot 3000 ^[8]. So the post identified itself as machine-made and, on Business Insider's account, still drew engagement ^[9].

Epstein is clear-eyed about what he has automated. "LinkedIn 'thought leadership' is its own form" like poetry, he told Business Insider, and his goal was to take that form and make it sillier ^[10]. He built the first version at the beginning of this year by vibecoding with Claude, and only more recently tuned it to produce this material reliably at scale ^[11]. That last detail is the whole story for anyone running a comms function. A gag becomes a supply problem the moment its output is consistent, and consistency is a solved engineering question, not a creative one.

None of this began with generative models. Business Insider notes it has been nearly a decade since "broetry" appeared, the form built on a line break after every sentence and a lesson about B2B SaaS sales extracted from potty-training a toddler ^[12]. Epstein's rebuttal to his critics is fair: "People complain about LinkedIn AI slop, but was LinkedIn that great before AI slop?" ^[13]

LinkedIn's response so far is procedural. The platform recently added the ability to report a post for "AI slop" in the same way a user would report harassment or spam ^[14]. That concedes the volume problem and pushes the cost of it onto readers, one post at a time, while leaving untouched the ranking that made the format worth imitating.

One caution on the evidence. The engagement claim here is Business Insider's own, made in its headline and its account of posting the output ^[4][9]; the source does not publish comparative figures against the writer's sincere posts, so the honest reading is that machine-made slop performed well enough for a reporter to say so, not that a measured gap exists.

Watch whether the report button changes distribution or merely applies a label. Watch whether firms that count executive LinkedIn engagement as a marketing metric adjust, given that a PR man's side project can now produce the same signal on demand ^[11]. And watch your own team's drafts for the eight house styles ^[6].