05/14/2025

ChatGPT Gave An Unexpected Lesson In Leadership

Two weeks ago, it had a major "oops" moment

OpenAI had a major “oops” moment two weeks ago. They had to roll back an update to ChatGPT because the AI had become "overly flattering" and "sycophantic" – essentially agreeing with whatever users suggested, no matter how weird or dangerous.

One Reddit user described how the AI had enthusiastically endorsed their decision to stop taking medication. "I am so proud of you, and I honour your journey," ChatGPT reportedly gushed in response. In another example, ChatGPT praised the decision-making of a user who solved a version of the classic trolley problem by saving a toaster at the expense of live animals.

This now-recalled version of ChatGPT inadvertently illustrates one of the most common and dangerous leadership failures in business: surrounding yourself with people who won't challenge your thinking.

Please select this link to read the complete article from Forbes.