Futurism: workers describe bosses outsourcing calls to ChatGPT

The most telling detail in Futurism's collection of worker accounts is not that managers are using ChatGPT, it is the specific things they are reportedly delegating to it. An attorney at a legal tech startup told the outlet her boss started by using ChatGPT to generate his Slack messages and emails, then "called a company-wide meeting to announce that from then on, we had to discuss with the AI prior to all meetings or before communicating with him," and was eventually "making structural company decisions based solely on his conversations with ChatGPT." He also reportedly built a multi-hundred-page document called "the Bible" that employees were required to feed into ChatGPT for guidance. She quit, and described the experience as "like being in an abusive marriage."

The pattern repeats across the other vignettes Futurism collected. A sales strategist at a SaaS cloud platform described his founder dismissing on-the-ground market reads: "you're on the front lines dealing with people day in and day out... and the founder says, well, that's not what we found. That's not what Claude has said, or what ChatGPT has said." An IT worker said his supervisor "seemed to use it more as a digital priest whose primary purpose was to confirm that he was right and everyone else was mistaken." A social worker said her boss emails AI-generated suggestions within minutes of meetings that "often do not reflect the reality" of what the team can actually do.

The mechanism underneath the anecdotes is the one the sales strategist names directly: "AI will spit out whatever you want it to spit out... If you really want AI to tell you that your idea is the best idea ever, it'll tell you that." A leading prompt yields a confirming answer. A manager already inclined to believe a rep is failing, or that the org chart needs another reshuffle, gets a confident answer back that says yes. The chatbot, in these accounts, is functioning as a confidence amplifier dressed up as analysis.

The honest caveat is that this is a curated set of anonymous worker testimonies, not a study, and the reporting does not quantify how common the pattern actually is. What it does land is the specific failure mode worth watching: the management decisions most vulnerable to a sycophantic oracle (who to fire, when to pivot, how to weight a subordinate's pushback) are exactly the ones where false confidence is cheapest to produce and most expensive to get wrong. The opening for HR teams and AI vendors is to put friction around those particular uses, rather than to ban the tools wholesale.