Andy Warhol at the Bruce Museum in Greenwich:

I live on the NY/CT line. I cross state borders to get groceries. So I have to ask: will my AI get smarter when I cross the road into Connecticut? New York's S7263 would make companies liable when a chatbot impersonates a licensed professional — a doctor, lawyer, financial advisor — and causes harm. The intent is sound. AI giving bad advice while cosplaying as an expert is a real problem worth solving. But software doesn't stop at state lines. And that creates a real tension. If every state writes its own rules, companies face an impossible choice: build 50 different versions, or default to the most restrictive one for everyone. Either way, users lose. Here's what keeps me up at night: the people most exposed to bad AI advice are often the same people who can't afford a real doctor, lawyer, or advisor. They're not using AI recklessly — they're using it because it's the only option they have. Over-restrict, and you've just taken that option away. Getting AI governance right isn't just about liability. It requires: - A clear line between advice and information - Consistent disclosure so users know what they're dealing with - Frameworks built to scale nationally, not state by state The groceries are the same on both sides of the border. The AI shouldn't be different either. Curious how others are thinking about this.