Congress Says It's Not Writing the AI Rules. Your State Will.

Speaker Mike Johnson went on CNN Sunday morning and said Congress isn't going to lead on AI safety. He pushed back hard on the idea of an emergency moratorium, said the responsibility sits with the companies building this stuff, and floated getting the AI CEOs into a room at the White House to work out guardrails among themselves. He also brought up China, which is the usual reason given for not slowing anything down.

That landed at the end of a strange week. Anthropic's CEO published a long piece Friday asking his own industry to pace itself. Sam Altman told Fortune that OpenAI won't go public this year because the safety work isn't done. So the people building the tools are saying take it slow, and Washington is saying that's on you, figure it out.

I don't have a side to sell you here. For a shop in Toledo the argument matters a lot less than the result, and the result is easy to read. Nothing federal is coming any time soon. The rules that eventually apply to the AI in your business are getting written somewhere else.

They're already getting written

Look at what happened three days before that interview. Newsom signed a pile of AI bills in California on the 10th. One of them puts real requirements on anybody operating a chatbot that talks to minors: time limits, mental health resources built in, a published safety plan, a parent alert if the thing picks up on self harm. Another makes platforms drop infinite scroll and autoplay for anybody under sixteen. There's even a hold on AI toys for little kids in there.

Colorado's AI act has new draft rules landing later this month. Texas, Illinois and Utah have all passed something in the last couple years.

Ohio hasn't done much. Michigan hasn't either. I'd treat that as where we are right now and not as a promise. State bills move fast when a story breaks, and the ones that pass usually don't carve out an exception for businesses your size. The California chatbot rules apply to whoever is operating a chatbot. That's it. That's the test.

Why a California law reaches a Toledo shop

Most owners I talk to figure state AI law is somebody else's problem. The reason it isn't is that your customers aren't all in Ohio.

If there's a chat widget on your website, it answers whoever shows up, from wherever they're sitting. Half the businesses I work with around here already serve Monroe and Lenawee County, so you're crossing a state line before lunch. When a rule attaches to the person on the other end of the conversation instead of to the company, your address stops deciding which rules you're under.

Three things to write down this week

None of this needs a lawyer yet.

1: Get an actual list of what's running. Every AI thing touching your website, your phones, or your inbox, on one page, with what it does and who set it up. You can't follow a rule about your chatbot if you forgot the web guy installed one in 2024.

2: Figure out whether any of it talks to kids. Almost everything moving at the state level right now is aimed at minors. If you run a gym, a dance studio, a tutoring center, a pediatric office, a youth league, your chatbot is having conversations with fourteen year olds and I'd assume you get regulated before the machine shop does.

3: Put your own rules on paper. What your people are allowed to feed into an AI tool, what they aren't, and who reads it before anything with a customer's name on it goes out the door. I wrote about this a month ago and I'm going to keep bringing it up, because one page of your own rules beats sitting around waiting for somebody else's.

Johnson could change his position next month. Congress does that. But if you're holding off on deciding how AI works in your shop until Washington tells you what's allowed, you're going to be holding off for a long time, and the first rule you actually run into won't come from Congress anyway. It'll come in a letter from somebody's attorney.