New York could be on the verge of a milestone AI safety bill

When California recently passed a first-in-the-nation law to regulate catastrophic risk in frontier AI models, New York was also in striking distance of the same distinction.

The Empire State’s legislature passed a similar bill, called the RAISE Act, in June, which has been waiting on a signature from the governor ever since. The bill would require large frontier AI developers to publish safety plans and prohibit the release of models with catastrophic risk—i.e., mass casualties or billion-dollar damages.

Alex Bores, a Democratic assembly member who co-authored the bill, told us Governor Kathy Hochul has until roughly the end of the year to act on it. VC firms and tech industry groups are currently lobbying against it.

Meanwhile, Bores has become one of a growing number of Democrats—nine and counting—to join the race for New York’s 12th Congressional District in mid- and upper Manhattan. He thinks AI policy will help him stand out from the pack.

But it’s also put him in the tech industry’s cross-hairs. Leading the Future, a $100-million super PAC funded by various venture capital firms and AI companies, told Politico this week (after our conversation) that it would take aim at Bores’ NY-12 campaign as its first target.

We spoke with Bores about the role of state legislation in AI, his background in tech, and how he’s making AI policy part of his campaign pitch.

This conversation has been edited for length and clarity.

How are you feeling about where the RAISE Act is at right now?

I feel good. It is a big bill that is taking on a pretty meaty issue, and so we’re having ongoing conversations about it. New York has this unique process called chapter amendments, where we can actually amend the bill. The governor isn’t just in a binary choice of signing or vetoing, and I expect we’ll begin the formal process of potential amendments soon.

The RAISE Act and California SB 53 that passed recently both focus on this idea of catastrophic harm. Can you talk a little bit about why you decided to focus on that specific level of risk, and if there’s a strategy there around focusing on these really extreme cases?

There is a lot that we have to do in AI, and the RAISE Act is just handling part of that. I have other bills that focus on deepfakes or on data transparency, or my colleagues have bills on the workforce…This is just one of many bills that I and others are doing, but it’s addressing an issue that is not really being addressed elsewhere, which is that this research is extremely powerful, and that’s a good thing. It can bring forward innovations that we’re really excited about, but if misused or done unsafely, could lead to some very significant harms. And so after seeing the federal government take no action on this front, you’re seeing a number of states say, ‘Well, someone needs to step up and protect our citizens.’ And so you’ve seen California pass a bill. New York obviously passed the bill, but you’ve seen efforts similarly in Illinois and in Massachusetts and in Vermont and in Colorado. You know, a number of state legislators are saying, someone needs to take action here.

Some venture capital firms and tech companies are spending big to try to block this. Did you expect that level of opposition?

Yes. And to be clear, I worked really closely with a lot of people in industry to get the details right. I did a first draft of this bill back in August of 2024 and I sent it around to all of the major developers, and I got red lines and feedback, and then I produced another version in December, and I got another round of red lines. We published the first version of the bill in March. I got more red lines. We amended it in May. We amended it in June. And so what you’ve seen is a wide swath of support, including from some VC firms, from some New York-based startups, and certainly from a lot of the people that actually work in AI day to day. Two of the godfathers of AI, a lot of academics that study it, support the details of the bill. But yes, I knew there would always be a set who are ideologically opposed to any regulation whatsoever, even if it’s regulation as light-touch as saying, ‘You have to design your own safety plan and stick to it.’

You come from a tech background. I think you said that you were the first person elected to the state legislature with a computer science degree?

I’m the first Democrat in New York at any level with a degree in computer science.

How did that prepare you for dealing with AI? Did you feel like you have more of a handle on it because of that?

To start off, it meant I wasn’t intimidated in diving into the issues. I have a master’s degree in computer science with a focus on machine learning. And more than that, I worked in industry for almost a decade, including at an AI startup that worked with early versions of transformers, the big discovery that has led to this surge in generative AI. And so these are issues I was familiar with. I’ve seen from both a technical standpoint and an industry standpoint, the incentives that shape the behavior of companies. At its best, the legislature has people with many different backgrounds who are coming together to work on issues. And I have this, for a state legislature, fairly unique background in technology, and so I do a little more on these issues than others.

When you were working on transformers very early on, did you imagine that they would eventually evolve into something like LLMs and this future that we have now?

I did not think it would come as quickly as it did. The products that I was working with were not that great. That company ended up going under.

How much of a role is AI policy playing in your current campaign for NY-12?

A lot. I mean, there’s many, many issues we need to deal with. So I’m going to be talking a lot about infrastructure funding in the district and the minimum wage and healthcare and education broadly. But also, I think AI is one of the issues that Congress is really missing the ball on. And we don’t want to be in the same situation we find ourselves in with social media, where it’s 10–15 years after the fact—20 years after the fact—and we’re trying to now graft on policies.

What you saw from those who have looked at this issue, for example, a California report that Governor Newsom impaneled after he vetoed 1047 said we have a limited time window where governments can act and really shape policy meaningfully, because if it goes too far, A), you don’t know what world we’re in, but B), practices might be so ingrained that it actually becomes harder to shape this in ways that benefit everyone. And so, yes, I think one of the things that is missing most in Congress is people with both the background in this field and the willingness to push through difficult bills that make things better for everyone.

It seems like there’s not much hope for a big, comprehensive AI regulation bill right now. But maybe they’re thinking about how you can regulate issues that everybody agrees on, like certain deepfake issues or child safety. Maybe you think that a big comprehensive push is more possible, but what are your priority issues?

I think it’s more possible than people think. I mean, your reporting…about how the majority of both Democrats and Republicans are worried about AI shows a quickly shifting landscape. But yeah, I don’t necessarily think that the right way to go about this is one comprehensive bill. It’d be like saying, ‘Let’s have one economy bill.’ It is so big and touching so many aspects and critically, moving so quickly that yeah, I think it’s going to take a lot of different steps forward.

And there’s work we have to do on regulating chatbots to protect children. There’s actually really exciting developments, technically, on how to solve deepfakes, and with a little bit of policy push, I think we’re actually much closer to a solution there than most people are giving us credit for. And then there’s a lot we need to think about in the education space as we continue to learn more. All of these can move in tandem and separately. If we try to bundle up to one big, controversial, universal bill, I worry that we end up missing the boat on many issues.

With a lot of different states acting on AI, the tech industry has argued that it creates a patchwork of different regulations. Do you think that’s a real problem?

Yes, and what I mean by that is, for many of these aspects, the best solution would be a federal bill. But the federal government moves slowly, and so even stepping outside of AI, states have always had this role of the laboratories of democracy, where states will try out new things. You can see the impact that it’s having, and then eventually the bad policies fade away, and the best ones might become federal. And with something moving as quickly as AI, that unique role of the state stands out.

I think a lot of federal lobbyists are very worried about action at the state level because they think states are as static as the federal government is. In 2023, Congress as a whole passed 27 laws. New York State passed 774, and we regularly sunset laws and renew them and change them. With the RAISE Act, if that is the only standard five years from now, 10 years from now, I would say that’s a failure. I would hope that we learn and improve it and eventually have a federal standard from there. So to get back to your original question, if 50 states have 50 different rules forever, that is not good, but having some action at the state level and having us learn, and then having that drive a federal conversation is the natural way that we should be doing things.

With all the candidates in this New York 12 race, do you think that AI helps you kind of stand out from the pack here?

Yeah, absolutely. Individual members of Congress, especially in a district like NY-12, a lot of us are going to say the same things on many issues. We’re probably going to vote in pretty similar ways. It is extremely unlikely that any of us are the deciding vote on a big piece of policy that we disagree upon.

But individual members—if they bring unique experience, a unique background, a unique expertise and history of moving things—can maybe introduce bills or push in certain ways that wouldn’t have happened without them. And so while I think a lot of us will have very similar voting records, I am making the case, and we’re getting a lot of positive response to the fact that if I end up being the congress member for the 12th District, we will see different policy around AI, we’ll see a different focus on AI, we’ll see more energy around solving these problems than we are currently seeing in Congress broadly, and that’s a distinct and unique reason to support the candidacy.