Yvette Schmitter with Fusion Collective on How Lobbying Is Weakening AI Regulation Before It Becomes Law

Hello, and welcome to the latest edition of the Techstrong AI Leadership Insight series. I'm your host, Mike Azar. Today we're with Yvette Schmid, who's the CEO for Fusion Collective, a consulting firm.

And we're talking about, uh, AI regulations, and specifically what's happening in New York state and how it might also pan out elsewhere, and maybe the differences between our friends in the state capitals and the nation's capital. But Yvette, welcome to s Shah. Thanks so much, Mike, for having me.

All right. So New York State passed this RAISE act, which was signed into law by the governor. And, um, like most laws, there's a little bit of a difference between what ultimately was agreed upon and may be enacted.

But for those of us who don't live in New York and or who may not have been paying attention to that matter, what is the RAISE act? What does it do and, and why, why should people pay attention to what's happening here? Well, that is a great question.

So let me walk you through what kind of happened with the New York State Raise Act, uh, what was passed by the New York Senate, and then ultimately what was actually signed into law in December by the governor of this wonderful state of New York. So in June of 2025, the legislative body of New York State, they passed what I would call a very people protective forward AI regulation that protected the people. It had mechanisms in it around, um, mandatory independent audits.

You know, when you do self-auditing, you're kind of like grading your own homework on that sense. So this act in June, it had external independent audits, um, to verify the claims that all these AI companies are making. It had comprehensive whistleblower protection.

So if someone was working at an organization that was seeing things that didn't kinda look right, they could report and be and be protected. There was penalties for wrongdoers that were up into the millions, like tens of millions for the first violation, which would actually probably deter bad behavior, you know, and in order to be considered or covered for, um, this act, it was based on compute thresholds. So you can't gain the system.

Now that passed, I was super happy in June. I was like, wow, this is gonna be amazing. But I kept reaching out to the New York State Senator.

I was like, well, well, when is this thing gonna get signed? Fast forward six months later, the governor of New York, she, what she signed and what was passed completely different. So think about independent audits gone.

So now it's being replaced with self-certification, whistleblower protections, completely gutted, not there, penalties slashed by 90%. So for the first violation, and here's what really gets me the compute thresholds. So it's hard to game compute because it's like electricity.

What you spent the, the, the, the tokens, you can't really game that. But you know what they did? Instead of having compute thresholds, they based it on revenue.

So now what you can do is create these corporate loopholes to restructure your organization. So now that your AI system that you've been developing, deploying doesn't meet the thresholds to be, to be oversaw by the RAISE act. So, and this is not my interpretation, this is not the world of ut, this is a side by side.

You look with what's passed in June and you look what the governor signed in December. And of course, none of this doesn't take place until January or 2027. It's completely different.

It's completely different. S how did that come about? And was there some sort of, was the act discussed again in the Senate somewhere?

In, in the No, No, no, no. This is where our lobbyists, our tech lobbyists, you know, tech, NYC, you know, Andreesen Horowitz, all the tech lobbyists, they got in and basically they gutted the regulation in order for it to sign. And so what the governor and the, the state legislators trying to say now is that the New York Raise Act is like nation leading.

This is not nation leading. It's not nation leading at all. I mean, basically what it does is it gives the, the 10 frontier model companies meta, Google, OpenAI, anthropic, it gives them a pass.

And, and it was touted as, oh, no, we can't have all this regulation because it's gonna, it's gonna stop innovation. It's not gonna stop innovation. Really.

It's doing, it's giving people a pass to do not the right thing. But now they gotta check the box saying, we did this, we did this, we did this, we did this, and it really does nothing. And guess who pays for it?

Us New Yorkers, the people who have to live with these systems that are commercializing the products. What people don't understand is thatAnd, New York City is literally like a AI startup hotbed. There's over 2000 AI startups here in New York City alone, 2000.

They're doing everything from FinTech, healthcare, you name it. They have AI in it, they got some funding, and they're here in New York. And I actually went to one of these startup meetings where they were doing pitch pitch decks, and they're like pitching their new ideas.

And I asked five FinTech startups, I said like, um, you know, how are you thinking about ethics? They looked at me like I had six heads. They're not thinking about ethics.

They're not thinking about how their AI system that they're gonna commercialize and then launch onto us unsuspecting New Yorkers who are gonna use these products, the impacts that it's going to have. So it's, it's, it's, it's sad. It's sad.

So, so what should people be looking out for? Or what should they be on guard for with these applications that are being rolled out, many of which are consumer oriented and financial services oriented? And are there things that are gonna happen that we're all gonna wake up in the morning and say, wait, who let this happen?

And maybe there'll be a backlash that, You know, um, Mike, there's, I say this all the time, if you wanna catch a squirrel and raccoon you shiny objects, but you don't do that to run a business. And I think how the organizations today that are getting this massive funding and how everything out there in our, at our fingertips has something to do with ai. You can put your, you can make yourself a storm trooper.

You can recreate a Star Wars theme. You can, you can see squirrels riding yachts and flying planes. You can do all these things, shiny objects, but no one's thinking about the cost of that.

When you want to turn yourself into an AI action hero, are you thinking about the power that it takes to run the algorithm and the compute and electricity to generate, you know what I mean? Like, no one's thinking, no one's shifting left. No one's understanding that this takes an immense amount of power that's coming from a data center that's sitting on a grid, that is a Tinder box at best.

You know, that needs to be cooled. Where you gonna get the water from? Who, who's paying for the extra power being, being charged from the electric, from the electrical grid?

Electricity is a, is a public re it's not private. You just can't throw a data center in someone's backyard and flip the switch and then all of a sudden no one, you know, who's gonna pay for the extra stress on the system? No one's really thinking that through, but they're looking at the squirrel videos and they're turning themself into AI action heroes.

I think we, we as consumers really need to think about what does this cost me personally? No one's reading terms and conditions. So Mike, I'm gonna ask you, when a new platform comes out and you hear about it and someone's saying about it, what do you do?

You scroll right through the end, right? And you hit click, click accept. Do you read, you know, like the nine pages of a terms of service?

No, no, it does, because they wanna, they wanna use the glam box. They wanna, they wanna use the app. They're not gonna read people that, wait, Wait, wait, there, there, there was a document with terms of service in it.

Where, where was that? You know, like no one, no, people blow past that. And that's where if you read that, you probably wouldn't sign, you probably wouldn't hit accept because mm-hmm.

That's where all of the details about what you're giving up as an individual. And I, I like to say this, I like to say we have become, we have gone from being customers to a data point. We're no longer customers.

You know, when you go into a store, like the customer's always, right? Okay, I don't wanna use that. Like what?

You go into a store they like, they actually care about your experience and how you felt and what you bought and the quality of the service you receive, that's gone. We are, we are just consumers. And so since we're consumers, that whole element of reading the terms and conditions, no one wants to read the terms and conditions.

Why? 'cause they just wanna use whatever, whatever product you're hawking today. So do you think that this will become a real political campaign issue as we go forward and people will start voting based on these AI issues?

Because to your point, it's not just my electric bill, of course, but it's also crazy things like, you know, grok is letting you take pictures of people and, you know, distribute you accordingly. And, you know, this is kinda not the kind of society maybe we wanna live in. You know, Mike, I I, I don't know if it's gonna become like a political issue because the, the folks down in DC have literally are, are making moves to make sure that no state law can impede AI innovation, right?

All this is being tucked under innovation. Um, but when I tell you what I'm about to tell you, this should really elevate for you. And as, as your listeners here, um, so in the California state law, they have a similar AI bill, uh, to New York.

They have the enforcement act of the law about figuring out if someone did something wrong. It doesn't happen until 50 people die. 50 people that's in California.

So that, that means if your AI system, your product, your impacts 49 people adversely or 49 people die, like there nothing happens. You can just keep running. This is per AI system in New York.

It's a hundred, a hundred people will have to die in order for the enforcement provisions of the raise act to go into effect. So, so let me just break that down to you. There, it kind of shows you like who's important, right?

Like if 75 people die, because an AI system, this is per system, right? So that's, your system does something unlive, 49 people, nothing happens. Nothing happens.

Another system does something under a hundred people, nothing happens. So the compounding effect thousands of people can be, can be unli and have adverse reactions. And based on the laws on the book in New York and California, nothing will happen until that one system impacts a hundred people in New York State.

So a hundred people have to be unli before the enforcement element of the raise act to go into effect a hundred people. So now tell that to the other 75 people who have to plan funerals, right? I mean, when you think about that, like there's no, there's no, there's no recourse for those, for those people.

And I think that is what that, I think that's what's gonna move the needle when, when adverse actions start happening, and then people look to their legislature in the laws and say, go after them legally they can't. Mm-hmm. From an enforcement perspective, I think that is going to move the needle.

And to add insult to injury, Mike, guess who the regulatory body in the New York state and New York state is for the, the Raise Act and the, uh, large frontier models and developers that fall under that. I have no idea. Department of Finance.

Hmm. So the bank, so the group who regulates banking is the enforcement agent for large frontier models who fall on the raise the Raises Act here in New York. What do they know about that?

Let me ask you something too though about, well, you know, let's take these incidents where there might be a fatality, but who's to determine what the cause of death was? I mean, was it the, the self-driving car? Was it a car accident?

Or did the AI do it? Because, you know, at that point, if I classify it somewhat differently, then I'm under the threshold, right? Yes.

And this is where gaming the system comes out. Because if you go on compute Mike, you can't gain compute, but you can restructure your organization. And I'm gonna give you an example.

So, um, let's use, let's use, let's use meta, let's use meta. So you have the mothership, they can spin off or, or, or think more about like, like pharma. Like pharma, uh, Pfizer, they have an r and d arm, which is separate and exempt, but the r and d arm is funded by the, the corporate, the parent company, right?

But they're separate. But if you put the r and d for your model under a separate entity, it doesn't meet the threshold. If your, if your revenue is under 500 million, that's how you can gain the system in New York, because the New York is, you have to have 500 million annual revenue plus your model size in order to, to be considered a covered entity under the raise act.

But you can do some corporate restructuring where, oh, I can do all my development under this other separate entity. It doesn't meet the revenue threshold, so I'm exempt. So I think that's where people, people, that's how you can gain the system.

And, and then the onus is gonna be on the per on the family, on who's left over on who was harmed to figure out who shot John and who to sue. But based on the raise act, as long as they make under 500 million, they don't harm a hundred people. They're not regulated.

And that entity might be registered in Delaware anyway, so it might not even be a normal New York kind of issue. Um, is this gonna play out in every state? I mean, are we gonna see the same effects?

Somebody's gonna, you know, pass a law and then we're gonna see this play out again and again where lobbyists come in behind the law and or that was at least as it was passed, and before it's enacted, there's gonna be all this, you know, activity in the, what did we used to call this? The, the, the smoke filled room. So, um, there, I, the LA at my last count, Mo there's like 19 or 20 states in addition to, uh, including California and New York who passed around things that are passing bills and legislation that is AI adjacent, you know, they're banking things under, like data, you know, data for data brokers and how you can use people's data and consumer protections.

They're not coming out and saying like straight, like ai AI system laws and, uh, rules and regulations. They're doing that. They're also, um, everyone is kind of like on the clock with the executive order that came out from our current president about the moratorium on AI laws on the state.

And so if the federal government determines that your law, potential law is gonna be discriminatory or, or mentions anything about discriminatory, they will shut you down and and sue you. So you can't, so that that law will never be enacted under the guise of ai. And we need to innovate.

So I do think, like the challenge is gonna be, you got states who don't want the federal government coming in, blowing stuff up, right? Under the guise of AI innovation, right? So they're gonna, they're gonna be creative, maybe do things on the privacy and, and, and under data protection and consumer protection, maybe that route.

And then you have the federal government where it's put the moratorium on states. It's like you can't pass any law that it impacts AI on their innovation, or you don't get federal funding for your state. That's, that is kind of like the, the, the, I won't say the networking, but the horse trading going on behind rooms.

Mm-hmm. That's the horse trading going on. If you want federal funding, you can't do this around ai, you know, and I think it's gonna be hard for states to, to stay out of the AI legislation race, but it's also gonna be hard for them to really get in and do the protections that's needed for us, the citizenry of their, of their states.

So do I what, call up my local congressman to express my concerns in outrage, or is this gonna have to be like, everybody's gonna have to take to the streets to force the conversation? I think it's gonna be a little bit of both. I mean, I mean, not to get super political, but just think about the, the Affordable Care Acts.

You know, like, hun, millions of people will not be able to afford health insurance anymore. Millions. They've, they, they're calling their, they're calling their senators, they're, they're calling everyone.

And it's, and that's still kind of not working. So I, I, I, I don't know. I mean, honestly, Mike, I mean, I don't know.

I think we as a people need to figure out, you know, what's more important, you know, and, and, and, and it's not about red state, blue state, Republican, democrat, inspirational person, whatever, whatever, whatever, whatever, um, uh, class you, political class you wanna put yourself or label you wanna put on yourself. At the end of the day, it's like, is this the world that we wanna live in? There you have it.

Hey folks, whether it's a law or a end user licensing agreement, by all means, don't forget to remember. Try that again. Hey, folks, regardless of whether it's a law or the licensing agreement for a piece of software, be sure to read the fine print.

Yvette, thanks for being on the show. Thanks, Mike. And thank you all for watching the latest episode of the Textron AI Leadership Series.

You can find this episode and others on our website. Please check them all out. Until then, we'll see you next time.