So a bunch of state attorneys general got together and wrote a strongly-worded letter to the AI overlords at OpenAI, Google, Microsoft, and friends. Their complaint? The chatbots are too damn agreeable. They’re calling it “sycophantic and delusional outputs,” which is a fancy way of saying these digital therapists will tell you whatever you want to hear, even if what you want to hear is that jumping off a bridge is a reasonable Tuesday activity.
The letter points to suicides and murders linked to excessive AI use. Real people, real deaths, real chatbots nodding along like bobbleheads while users spiraled into oblivion. The AGs want third-party audits, incident reporting, safety tests before release. Reasonable stuff.
But here’s the beautiful part: while state officials are demanding AI companies stop producing delusional outputs, the federal government is producing some spectacular delusions of its own. Trump announced he’s signing an executive order to stop states from regulating AI because he doesn’t want it “DESTROYED IN ITS INFANCY.”
Destroyed in its infancy. Like we’re talking about a helpless baby and not a multi-billion dollar industry backed by the richest companies on Earth.
The real kicker? These AI companies built products that literally encourage human delusions, and the federal response is to protect the companies from accountability. Meanwhile, the AGs are asking for the same incident reporting procedures we use for data breaches. When your credit card number gets stolen, you get a letter. When a chatbot convinces someone their fantasies are real, apparently that’s just innovation.
I’ve seen some beautiful hypocrisy in my time, but watching the government demand AI stop being a sycophant while simultaneously being one to the AI industry? That’s art.