ChatGPT is keeping receipts.
In January 2026, a federal judge ordered OpenAI to hand over 20 million de-identified
ChatGPT conversations as evidence in the New York Times copyright case. The users? Not notified. No chance to object.
That ruling didn't make headlines in HR circles. It should have.
Every prompt your managers type into ChatGPT or Claude is now potentially discoverable in litigation.
Not "in theory." In practice. Courts are already ordering it produced. There is no AI privilege. There is no "but I was just thinking out loud." There is the prompt, and there is the lawyer who will read it back to your manager on the stand. I first heard about this in family court, but it is definitely prevalent in corporate cases and likely more costly.
And founders, your "let me sparring-partner this with Claude" sessions about co-founder drama, board politics, and who to push out? Same rules apply.
We use AI too, as should you. We're telling you the cost of unstructured AI use just got real, and most of your managers have no idea.
Here's What's Getting Companies Sued, Fined, or Reputationally Damaged Right Now
The Fix Isn't Fear. It's One Add and Three Filters.