Insurance company Nippon Life just sued OpenAI for $10.3 million

A disability claimant uploaded correspondence from her lawyer into ChatGPT. The AI validated her concerns about the legal advice she'd been given. She fired her lawyer, then used ChatGPT to try to reopen a case that had already been settled and dismissed.

When a judge denied that, she filed a new case and dozens of motions that Nippon says ChatGPT drafted. No licence. No accountability. The insurer had no idea until it was back in court.

This looks like a LegalTech story. It's InsurTech.

Because here's what actually happened: AI sat between an insurer and a claimant and changed the outcome of a dispute. The insurer spent significant time, resources and legal fees responding to filings generated by a chatbot that isn't admitted to practice law anywhere.

Yes, this happened in the US, where the litigation culture and legal system are very different from Europe. But the underlying dynamic is universal: general-purpose AI tools are already being used to challenge insurance decisions, and that trend doesn't stop at borders.

And this isn't an edge case anymore.

Millions of people already use ChatGPT, Gemini and similar tools to understand their policies, question claim denials and figure out what they're actually entitled to. Not because anyone built an AI tool for insurance, but because ChatGPT is good enough, available at midnight and free.

Dedicated insurance AI is coming too. And it's not stopping at personal lines.

Commercial insurance is where the complexity lives. Liability, D&O, cyber. It's also where AI advisory could do the most damage, or the most good, depending on which side of the table you're on.

The way advice is delivered is changing. The question is how far it goes. We're not far from AI that doesn't just answer questions but acts on your behalf: comparing coverage, initiating claims, even pushing back on settlements. Some of this is already being prototyped. The rest is a matter of time.

The rules on advice, distribution and liability exist. But they were written for humans and licensed firms. How do they apply when the advisor is a chatbot with no licence, no supervisor and no professional obligation? That's the question nobody has answered yet.

Nippon Life may have just filed the case that starts answering those questions.

The insurance industry spent a decade asking if it could become more digital. The better question now: what happens when AI becomes the distributor, the advisor and the advocate?