Ever have one of those moments online where you try to be helpful, armed with nothing but the actual facts, and you get slapped with a digital lifetime ban for your trouble? Welcome to the perplexing world of Perplexity, where citing their own manual might just be the ultimate crime.

A user over on the official Perplexity subreddit did exactly that. They—re a paying Pro subscriber who noticed the much-advertised —Deep Research— feature seemed— shallow. So, they did some research (irony alert) using Perplexity—s own official documentation and launch blog to prove the feature was severely throttled and not living up to the hype. The community agreed, pushing the post to the top of the sub with hundreds of upvotes. The company—s response? They deleted the thread and handed the user a permanent ban. Talk about shooting the messenger.

It—s the classic tech company two-step: advertise a shiny feature, quietly nerf it later, and then treat anyone who notices like they—ve leaked state secrets. The funniest part is they didn—t ban them for being wrong or rude—they banned them for being too right, and for using the company—s own words as evidence. It—s like getting kicked out of a book club for reading the book too carefully.

This whole saga is a perfect little internet culture capsule. It—s got everything: a disappointed customer, a viral community call-out, and a moderation team that chose the —delete and deflect— strategy over, you know, addressing the concern. The user even had to pin the evidence to their own profile after the purge. When your customer service strategy involves more cover-up than a paint store, you—ve taken a wrong turn.

In the end, it—s a witty reminder that in the land of AI, the most perplexing algorithm of all is still the one that decides corporate PR. Sometimes the deepest research just leads you straight to the banned list.