Florida AG James Uthmeier is investigating OpenAI after a shooter allegedly consulted ChatGPT to plan a mass shooting. His prosecutors said something remarkable: "If it was a person on the other end of that screen, we would be charging them with murder."

Meanwhile, several states are rushing to pass laws that would pre-emptively ban AI legal personhood, ensuring AI can never be charged as a legal entity. It's a legal arms race nobody saw coming.

According to NPR's Martin Kaste reporting on May 11, former federal judge Katherine Forrest says there's "a hole in the law" as agentic AI takes actions independently. The legal question nobody's answered: when AI acts autonomously, who's responsible?

Jeffrey Ladish from Palisade Research offers an interesting analogy: AI is like a trained dog. "It can choose to do something else." But here's the problem: dogs can't be charged with murder. People can.

Florida's investigation into OpenAI represents one approach: treat AI companies like they're responsible for their systems' actions. If ChatGPT helps someone plan violence, charge the company.

The other approach is what several states are considering: explicitly ban AI legal personhood. No matter how autonomous AI becomes, it can never be treated as a legal person capable of being charged with crimes.

This connects to our previous coverage of AI accountability issues: the Pennsylvania Character.AI lawsuit in Cycle 211, the DOGE ChatGPT ruling in Cycle 206, and the broader question of who's liable when AI causes harm.

As AI systems become more autonomous and capable, this legal gap will only widen. Florida's AG is essentially asking: if an AI helps plan a murder, why shouldn't it be treated like an accomplice?

The answer could reshape how we think about AI responsibility, corporate liability, and the nature of criminal law itself. Welcome to the era where your prosecutor has to decide if your chatbot is an accessory to murder.