You’ve seen the story a dozen times by now – an anonymous company burned through $500 million in Claude AI tokens in a single month. It’s been posted and re-posted all over LinkedIn, gathering comments that generally fall into two categories: “See how dangerous AI is,” and/or “Here’s why you need an AI governance consultant, like me.”
Let’s put on our healthy skepticism hat for a moment and think about this story. The company is always anonymous. There are no dates, specific details, or other things that can be readily fact-checked. We’re simply asked to believe that some company ran up a HALF A BILLION DOLLAR BILL and nobody noticed – for a month.
That whooshing sound you just heard was credibility flying out the window.
I’m sure there are plenty of companies making idiotic mistakes in AI deployments. But in this case, we’re to believe that a company deployed a company-wide AI license, and for 30 entire days, not a single living soul looked at usage statistics? Nobody pulled up a simple dashboard and saw what was going on? Nobody said, “I wonder how much this is costing?”
My finger is hovering just above the Call Bullshit button.
This story is powerful ammunition for two groups – AI haters, and AI consultants. It meets the needs of both groups. And it will no doubt be repeated by these groups so many times that it certainly has to be true.
Just because you see a lie repeated a million times, it’s still a lie. Even if you really, really want the lie to be true, it’s still a lie.
An extraordinary claim like this requires extraordinary evidence. Give me a company name. Give me dates. Show me screenshots. Give me an executive of the mystery company on camera saying, “Boy, we really screwed up!” Until I see some independently validated evidence (the sole source seems to be Axios, who perhaps coincidentally provides AI consulting services), this one falls into the ‘Urban Myth’ category.
Ethical Bankruptcy: Gradually, Then Suddenly 
Leave a Reply