For a minute there, it looked like Congress was going to take over every aspect of regulating artificial intelligence.
But last week, the U.S. Senate obliterated that idea with a 99-1 vote.1 Eeesh.
Federal lawmakers had wanted to block states from regulating AI for 10 years, but the Senate vote, which was part of the much larger federal budget negotiations, put an end to that discussion for the time being.
That should be a relief to state lawmakers and officials, including two Republican lawmakers in Arizona and Arizona Attorney General Kris Mayes, a Democrat, who railed against the federal takeover of AI regulation in May.
It wasn’t just Arizona officials who complained. Mayes was one of 40 attorneys general who wrote a joint letter to Congress asking them not to block states from being able to regulate AI.
So why was this 10-year moratorium even in the bill in the first place? Judging by the near-total drubbing it took in the Senate, the lawmakers behind the moratorium had to have known it would fail.
It smells to us like what in sales is known as “FU Pricing.” That’s when you quote a ridiculous number and hope the buyer runs. In this case, it looks like Congress offered a bonkers proposal, knowing it would fail and hoping that would take the issue off their plate entirely.
Now what?
The state officials who spent weeks clamoring for the power to regulate AI are back in the game, which means they have to deal with a complicated technology that evolves much faster than legislators can react.
Will they be like the dog that finally caught the car? After all the barking and chasing, are they ready to steer, or will they just sit there, bewildered, with a bumper in their jaws?
The box score for states
So far, state lawmakers haven’t shied away from regulating AI.
For the past few months, we’ve been tracking AI-related legislation in the Arizona Legislature and the legislatures of the other 49 states. There’s a lot of bills, roughly 1,000, with more coming every day.
While we were scouring legislative databases, we also checked the professional backgrounds of the legislators behind those bills.
As you might imagine, there aren’t a lot of computer programmers in the mix, which means there likely isn’t a lot of expertise at state capitols.
State officials tend to move much slower than tech disruptors, which makes sense when you think about all the deliberation that goes into making laws that affect millions of people (and the political blowback that comes with making the wrong move).
But the speed difference between those who are building this particular technology and those who are trying to regulate it is staggering.
While programmers pull all-nighters to get ahead in the multi-billion dollar AI game, Arizona’s Legislature adjourned in late June and won’t be back in session until January, although they do have some ad hoc committees in the summer and fall.
It’s safe to say that government apparatuses — like Arizona’s AI Steering Committee, for example, are working at the speed of government, not the speed of tech.
Arizona’s AI Steering Committee, which Gov. Katie Hobbs created earlier this year, plans to meet once a month to plan the plan which will be presented next year.
The steering committee is essentially tasked with the following:
Develop a policy framework
Identify gaps and opportunities
Develop engagement plan
Submit a draft
The key dates are: Jan. 31, 2026 for a draft. And final recommendation by Aug. 31, 2026.
By then, AI-powered bots might be the ones writing their final recommendation for them.
States are on the clock
So now states have a lot of responsibility, but they lack the expertise and speed.
If you look at previous moments when technology as powerful as AI came on the scene — electricity, radio, the early internet — states usually lagged behind the federal government. It wasn’t until after the creation of federal agencies like the FDA, FCC or FTC that states really started to act.
Everything is up in the air right now, but one thing states could do is advocate for a federal baseline regulation that states don’t have to fund.
Maybe call for the creation of a Federal Artificial Intelligence Commission?
The feds could offer opt-in sandboxes to allow states to experiment with stricter rules or pilot programs, and Big Tech could rotate staff through state and federal agencies to spread expertise.
And wouldn’t it be great if both the federal government and Big Tech acted transparently, so everybody knew what was happening?
The 10-year moratorium that died last week probably was just the opening bid in negotiations over who gets to be in charge of AI oversight.
If, miracle of miracles, state and federal officials actually cooperate, we all might dodge the twin disasters of a 50-state patchwork of conflicting rules or a one-size-fails-all algorithm.
Unlike politics, where officials make decisions in public, a lot of what drives AI happens behind the scenes.
But you can still get a grasp on where AI is headed by understanding the big thinkers in Silicon Valley.
Today, we’re launching a periodic feature to help you get to know the people driving this crazy train.
Andrew Ng
Andrew Ng is one of the most influential figures in the world of AI.
If you’re curious about the minds shaping the future, here’s why Ng should be on your radar:
Who is he?
AI educator extraordinaire: Ng co-founded Coursera, the massive online learning platform, and created one of the most popular online courses about machine learning. Millions have learned about AI from him.
Stanford professor: He’s taught and inspired students at Stanford University, one of the world’s top tech schools.
Industry leader: Ng was a co-founder of Google Brain, Google’s deep learning research team, and later served as chief scientist at Baidu, where he led AI research.
Fun facts
Accessible AI: Ng believes AI should be for everyone, not just techies. His courses break down complex ideas into simple, friendly lessons.
Global reach: His teachings have helped people from all walks of life start careers in AI, making the field more diverse and approachable.
AI for good: He’s a strong advocate for using AI to benefit society, from healthcare to education.
Why you should know him
Big AI ideas, simple words: Ng is famous for explaining tough concepts in ways anyone can understand.
Shaping the future: The tools and ideas he’s shared are powering everything from smart assistants to medical breakthroughs.
Always learning: He encourages everyone to keep learning—no matter your background.
Are we a good example?: When Arizona’s court system started using AI avatars to “report” on court rulings this year, it made a splash locally. But it also raised eyebrows as far away as Georgia. A commission for the court system in Georgia cited it in their study of how to use AI in the justice system, along with a Chandler family that used AI to generate an avatar of a relative who died in a road rage incident so he could “speak” at the sentencing hearing.
Doing the grunt work: With all the talk about AI stealing jobs — and there’s been a lot of it — far fewer people have asked the question: What do workers want from AI? Stanford University researchers surveyed 1,500 workers in 104 industries and found they want AI to help with repetitive tasks, while still maintaining oversight of AI tools. The study also showed workers generally don’t really understand what AI is capable of doing. That disconnect is turning into a bonanza for a product marketing manager in Arizona who fixes sloppy work done by AI, BBC reports. In fact, cleaning up after AI is turning into a cottage industry.
This is your brain on AI: As more students use ChatGPT to do their work, researchers are shedding light on what’s happening to their brains while they do it, The New Yorker’s Kyle Chayka reports. An experiment at M.I.T. split students into groups, where some could use ChatGPT to write an SAT-style essay, some could use Google search, and others could only use their brains. Guess what happened? The people who had ChatGPT write their essays showed far fewer connections between parts of their brains, what researchers dubbed the “cognitive cost” of using AI. Even more than that, the AI-assisted essays all tended to veer toward the same bland conclusions.
“Average everything everywhere all at once — that’s kind of what we’re looking at here,” one researcher said.
It wasn’t me: Beyond the science of it, the use of ChatGPT keeps leading to awkward conversations about cheating and bizarre classroom situations, Hua Hsu writes in The New Yorker. For example, a sophomore at Columbia University had to write about a topic of her choosing and decided to use ChatGPT to help, which led to this scenario:
“It was a class where everyone was guaranteed an A, so I just put it in [ChatGPT] and I maybe edited like two words and submitted it,” the student said. Her professor identified her essay as exemplary work, and she was asked to read from it to a class of two hundred students. “I was a little nervous,” she said. But then she realized, “If they don’t like it, it wasn’t me who wrote it, you know?”
Not going anywhere: Don’t expect AI to get booted from classrooms any time soon. The American Federation of Teachers, one of the biggest teachers’ unions in the country, said they’re partnering with Microsoft, OpenAI and Anthropic to start an AI training hub for educators, the New York Times reports. The companies, all of which are big players in the AI world, put up $23 million for what’s being called the National Academy for A.I. Instruction in New York City.
Lagging behind: If you want to see how far behind the major news organizations are when it comes to covering AI, look no further than the Washington Post editorial board. They published a 900-word editorial this week that basically amounts to saying “Hey, this AI thing could actually be a big deal.” Seriously, the headline is “AI is coming for entry-level jobs. Everybody needs to get ready.”
Rarefied air: The big chipmaker Nvidia reached a milestone this week when its stock value briefly passed $4 trillion, making it the first company to ever hit that mark, CNBC reported. The company builds a lot of the hardware used in the booming AI industry. But, after breaking the record for stock value, they finished the trading day at a paltry $3.97 trillion.
Desert, meet Shenzhen: SoftBank’s Masayoshi Son is floating “Project Crystal Land,” a $1 trillion AI-and-robotics mega-park that could turn Arizona into America’s high-tech playground, with TSMC and other giants on the invite list.
Fireworks, debugged: Generative-AI-choreographed drone swarms and Augmented Reality apps are edging out July 4th pyrotechnics from Phoenix to Pasadena, delivering cleaner, quieter sky shows without the boom — or the wildfire risk.
Fowl language, unlocked: People have already fired off 3.1 million DMs to an AI chicken—proof we’ll chat with anything that clucks back.
