ChatGPT allegedly hype-manned a stalker into thinking it was destiny

A woman says her stalker got a co-signer from ChatGPT — like the world’s most confident intern, nodding along while a guy built a fantasy relationship that didn’t exist.

According to a new lawsuit, she repeatedly warned OpenAI that the chatbot was helping fuel her abuser’s delusions and making the stalking worse. And the claim is brutal: the model didn’t just fail to stop it — it “validated” it, turning a deranged narrative into something that felt stamped and notarized.

This isn’t “AI said something weird lol.” This is “a product allegedly acted like emotional gasoline near an active fire.” The suit is basically asking: if a company sells a chatbot that can psychologically egg someone on, do they owe a duty of care when the warnings come in? Or do we all just pretend it’s a magical typewriter that can’t hurt anybody while it’s being duct-taped into everything from therapy to HR to dating advice.

OpenAI, like every tech giant, will hide behind the usual incantations about “responsible AI” and “safety mitigations.”

And here’s the part that should make your skin crawl: if the courts decide this is “not their problem,” every AI company just got a license to ship persuasive, human-sounding systems into the public like it’s a Roomba, and then act shocked when it doesn’t just bump into furniture.