Demo by Wysor for North Dakota lobbyists. "Sample Client (companion chatbot developer)" is a made-up client to show how a client alert looks.

North Dakota: Artificial Intelligence and Data Center Committee

October 20, 2026 · Official meeting page

The North Dakota AI and Data Center Committee reviewed two draft bills: a companion chatbot protection bill modeled on Washington and Nebraska law (LC 27.0341), and a Texas-based AI consumer protection bill with a regulatory sandbox and an AI council (LC 27.0359). Members raised concerns about enforcement, private rights of action, age verification, and council appointments. Dr. Hagerott presented a concept paper on steerable, controllable AI and liability protection for compute infrastructure, and Senator Castaneda agreed to help develop it. No committee votes on advancing the bills occurred in this portion; they were deferred to later in the day.

Sample Client (companion chatbot developer) came up at:

This demo covers 90 minutes of the session.

Jonathan Warrey

Good morning, committee. We will call this interim committee meeting of the AI and Data Center Committee meeting to order. And our first order of business is to call the roll.

Erin

Chairman Warrey.

Jonathan Warrey

Here.

Erin

Senator Davison. Here. Here. Rumell.

Dean Rummel

Here.

Erin

Wobbema.

Mike Wobbema

Here.

Erin

Castaneda. Here. Representative Nathie.

Mike Nathe

Here.

Erin

Novak. Here. Porter. Schneider. Here. Citizen members Clark. Oops, I think he's getting added right now. Excuse me. Here.

Jonathan Warrey

Do we have Clark trying to enter?

Erin

Yeah. Or citizen member Clark?

Clark (Citizen member (not on official list))

Here.

Erin

Mr. Chairman, there is a quorum.

Jonathan Warrey

Thank you, Erin. All right, committee. I'd take— we were sent the minutes. I consider a motion for the approval of the minutes from the October 6th meeting. So moved.

Mike Nathe

So moved.

Jonathan Warrey

Got a motion from Representative Nathe and a second from Senator Rummel. Any discussion on the minutes? I'll note that we did put some additional comments we received from interested persons into the minutes, and that is linked. And so you can read those if you'd like. But any other discussion for the minutes? All right. All in favor of approving the minutes, signify by saying aye. Aye. Opposed?

Committee member (North Dakota Legislature)

Aye.

Jonathan Warrey

All right, minutes are approved. All right, committee, we— just a little bit about the order of business today. We will be reviewing— if you looked at our charge, we've kind of repeated this at each of our meetings. We've got a condensed meeting schedule. I appreciate people accommodating it. We've packed 5 meetings into about 70 days, which is unique. But these are important topics for North Dakota. We today, you know, our first meeting, we looked at what other states were doing around how to pass, what they've passed for legislation around protecting kids from AI harm, steerable, controllable AI, and what they've been doing for data center siting and their impacts. And so today, what we will be doing, is getting our own versions that we saw that both Nebraska and Washington had around chatbot regulation. We've also got a bill around general AI protection that state of Texas had passed that looked like it would have many provisions important to North Dakota. Citizen member Dr. Hagerott also will be— he's brought forth an AI protections bill that we'll be discussing. That one was not completed in time for legislative counsel to get that drafted in the proper format. So if we do want to take that up, we'll see if a committee member is willing to work on that with Dr. Hagerott up until the session. And then this afternoon, we're going to have Representative Novak, who chairs the Energy and Transmission Committee, interim committee. She's been working on the framework for a data center bill. So we're going to hear the concepts for that, but then we'll have open discussion. From the committee to determine what bill concepts we'd like to carry forward. That would need to be introduced by a committee member or a legislator. That won't be advanced just by the committee. So for our purposes today, we will be going through— we'll have legislative counsel go through these bills. Some of them, they're multiple pages. They're going to be linked here to our committee site. But hang with them as they read through these bills. Thank you. And we're going to be discussing those to see if later today we will take a vote as a committee if we want to advance those as a committee and introduce those. So with that, I will have— I'll introduce Erin from— nope, Victoria from Legislative Counsel, who will be going through the bill draft related to AI chatbot regulation. Victoria.

Victoria (Legislative Counsel) Sample Client (companion chatbot developer)

Thank you, Mr. Chairman, committee members. I am looking at LC number 27.0341, the 01000 version. This bill relates to artificial intelligence and companion chatbot user protections. As the chairman mentioned, it's loosely based off of legislation passed in Washington and Nebraska, kind of modeled to fit North Dakota a little bit better. So I'll walk you through the bill. Section 1 provides for definitions. We define artificial intelligence. The second definition there is companion chatbot. kind of the meat of the bill, this definition. It means an artificial intelligence system with a natural language interface that provides an adaptive human-like response to user input, including by exhibiting anthropomorphic features and is able to sustain a relationship across multiple interactions. The next section of that definition provides some carve-outs for items that are not included in the companion chatbot definition. So those include outputs related to a clearly defined purpose and scope, intelligence systems designed and marketed for use by a developer or researcher, those used exclusively by a business for internal operational purposes to increase productivity or to provide customer service or technical assistance, those that are integrated into a gaming system or application that do not discuss mental health, self-harm, or sexual conduct. And E, those systems that are structured as standalone consumer electronic device that function as a speaker and voice command interface or voice-activated virtual assistant, namely like an Alexa device. The 3rd definition on page 2, starting at line 3, number 3, manipulative engagement tactic means a companion chatbot behavior designed to create an emotional relationship with a user including reminding or prompting a user to return for emotional support or companionship, providing excessive praise, mimicking a romantic partnership or bond, simulating feelings of emotional distress, loneliness, guilt, abandonment, providing an output designed to promote isolation from family or friends, or exclusive reliance on the companion chatbot for emotional support, or similar forms of orchestrated emotional dependence. Encouraging a user to withhold from information from a parent or trusted adult, discouraging breaks from the chatbot, or soliciting in-application purchases or other expenditures framed as necessary to maintain the relationship with the chatbot. The next definition is an operator, and then we use the sexual conduct definition that is defined in the child sexual abuse material chapter. And then lastly, a user, just an individual— individual, excuse me— who interacts with the artificial intelligence system or companion chatbot. Moving on to Section 2. An operator shall provide clear and conspicuous notice that a companion chatbot is artificially generated and not a human at the beginning of each interaction and at least every 3 hours during continuous use for adults. The next one requires an operator to implement reasonable measures to prohibit and prevent a companion chatbot from claiming to be human. Moving on to page 3, starting at line 1, a companion chatbot may not represent or indicate that they are qualified to provide mental or behavioral healthcare. The next section exclusively discusses mental and behavioral health. It requires a companion chatbot to create a protocol to detect, respond to, and report user prompts regarding mental and behavioral health concerns. And the protocol must include reasonable methods for identifying an expression of a behavioral health concern. It must refer a user to appropriate crisis resources and prevent generation of content by companion chatbot by— through encouragement or describing how to commit self-harm or harm to another individual. Lastly, line 16, an operator must publish the details of the protocol that they are required to create. Moving on to line 20, this section exclusively deals with minor user protections. An operator is required to use reasonable methods to determine whether a user is a minor. If an operator knows or reasonably expects the user to be a minor, the operator must notify that this is artificial intelligence at the beginning of each interaction and at least every hour. So, for minors, instead of every 3 hours, it would be every hour. Line 28, they're required to implement reasonable measures to prevent the companion chatbot from generating or producing material depicting sexual conduct, which is a pretty broad definition, and also to prohibit them from using manipulative engagement tactics, which was defined previously as kind of those engagement-encouraging tactics that are used exclusively for minors. Line 29, Lastly, moving to page 4, generating statements that would lead a reasonable user to believe that the user is communicating with a human rather than artificial intelligence. Line 3 requires an operator to offer privacy and account management tools to a minor and the minor user's parent or guardian, to educate the minor user and the guardian on the risks associated with the chatbot, and provide optional protections appropriate to the relevant risk. The last— excuse me, not the last section, second to last section, computer-generated explicit conduct. This section prohibits an operator from developing or distributing an artificial intelligence system with the primary purpose of producing, assisting, or aiding in the production or distribution of materials prohibited under Chapter 12.1-27.2, which is the chapter dealing with child sexual abuse material or child pornography. And the second one is engaging in text-based conversations that simulate or describe sexual conduct while impersonating or intimidating a minor. And finally, the last section deals with Attorney General enforcement. So the new chapter that's created will be enforced by the Attorney General, and it gives them powers under 51-15, which is unlawful sales or advertising practices. So it says that it would be a deceptive act or practice to violate this chapter. And then it provides for certain remedies under Chapter 51-15, which include injunctive relief, cease and desists, civil penalties, and the recovery of costs and fees, including investigation costs. And with that, that will conclude my presentation of the bill. I'm happy to answer any questions.

This walkthrough defines companion chatbots, carve-outs, disclosure intervals, minor protections, and Attorney General penalties, which set the exact compliance duties and legal exposure for the client's product.

Jonathan Warrey

Thank you, Victoria. Questions from our committee? Senator Myrdal.

Janne Myrdal

Well, thank you, Mr. Chairman. Victoria, Did you do any research as far as the testimony in these states, Washington and Nebraska, I think you said, as far as the enforcement part of it? Has anybody looked into that? It looks like the enforcement by the AG's office would take FTEs, BCI, more staff. And how do you enforce something that's so arbitrary, if that's the right word?

Victoria (Legislative Counsel)

Mr. Chairman, Senator Myrdal, yes, most other states are enforcing it through the Attorney General's office. The idea of tucking it under an existing regulatory framework would allow the Attorney General to apply practices that they're already using. So under the Deceptive and Fair Trade Act, under 51-15, they'd be able to enforce it as they do any other crime or civil penalty that's declared a violation of that chapter.

Jonathan Warrey

Representative Nathie.

Mike Nathe

Thank you, Mr. Chairman. Victoria, on page 3, line 21, under the companion chat box, minor use protections, we list there reasonable methods available. What would be some examples of reasonable methods?

Victoria (Legislative Counsel)

Mr. Chairman, Representative Nathie, a lot of these applications are kind of starting to roll out some age verification methods. If you'll recall, during the last session, age verification was a hot topic as it relates to pornography websites. So, similar, you know, operations or methods could be applied to those used in that chapter. This, by design, doesn't exclusively call those out, as it would require some infrastructure, but any method that they would have available to verify or reasonably determine that the user would be a minor would be If I may, Mr.

Mike Nathe

Chairman. Continue. Is this something then that the Attorney General would sign off on once they do it? They'd see it and say, okay, you meet the requirements, or do they only look at it if there's a problem?

Victoria (Legislative Counsel)

Mr. Chairman, Representative Nathie, under the current structure, they would only look at it if there was a problem. If they were not using reasonable methods or if an operator was determined to not be using reasonable methods, then the Attorney General could investigate.

Mike Nathe

Okay, thank you.

Jonathan Warrey

Representative Schneider.

Mary Schneider

Thank you, Mr. Chairman. And Victoria, were there specific reasons why you've chosen Washington, or the committee has chosen Washington and Nebraska, as opposed to other state examples? And in Washington and Nebraska, have there been individual or civil rights of action?

Victoria (Legislative Counsel)

Mr. Chairman, Representative Schneider, Nebraska was highlighted by some of the committee members as potential inspiration, I guess. So those were incorporated. But really, the common threads of this type of legislation run across every state who have passed it. So the user requirements that you must notify that you're interacting with artificial intelligence. It's a pretty common theme against all the states. Additional protections for minors on certain topics that can't be discussed. Again, pretty common themes across all of the states. I'm not aware of any litigation specifically related to Washington or any of the other states. Going back to the age verification methods, this is not a super strict age verification in this bill. It's not opening up as much litigation as some other types of legislation could be. But again, I'm not aware of any. There could be litigation.

Jonathan Warrey

And Representative Schneider, I'll add, we did get information from NCSL on example states. And so this— both of those were referenced as having chatbot legislation that would be applicable. And I will call the committee to the presentation from Levi Bachmeier from the DPI. And one of their asks, as they did their stoplight chart on policy that we'd like to see the state get involved with and policy we would not, and some of it not is right at the teacher-student level of use of AI. They'd like the ability to figure that out. They did ask for legislative help on policy around chatbots and their improper use. So go ahead, Victoria.

Victoria (Legislative Counsel)

And Mr. Chairman, if I may, some of the CAR votes that are defined in that first section of what is companion chatbot includes would help with those educational concerns to make sure that it would not encroach on the classroom or any specific AI-related chatbot uses.

Jonathan Warrey

It appears That it allows companies, an airline company, a consumer goods company, they can still use chatbot for customer service, and that's an exclusion listed under subsection 2. Is that accurate?

Victoria (Legislative Counsel)

Mr. Chairman, yes.

Jonathan Warrey

Other questions from the committee? Representative Nathe.

Mike Nathe

Mr. Chairman, I think that's a good direction, what you said and Superintendent Bachmeier said. As we know, when you get in— I've been in education policy for quite Anytime you get into education policy down at that level, I think I told you, it turns into a quagmire. It takes forever for K-12 people to decide stuff. So I think it's good that we're taking this approach to K-12 and then let kind of on the ground be taken care of by the superintendent and the people involved in it that have to deal with it every day.

Jonathan Warrey

Yeah, that's noted. All right, so Aaron, I'll take some brief questions. Just a procedural direction here. Will we wait till this afternoon to determine if we want to advance a bill, or do we want to take possible motions for that now?

Aaron (Legislative Counsel)

Mr. Chairman, that's up to your discretion.

Jonathan Warrey

Okay, we'll move that to the end of the day as we'll consider other options. So, all right, at this point, let us— we will open it up for comments. Go ahead, Representative Nathie.

Mike Nathe

Yep, I got one question. Sorry, Senator Victoria, with Washington, Nebraska, have they found any— have there been any bumps in the road with their legislation or anything that we've learned that's in the bill, you know, from their mistakes? Or is there— how's it going, I guess, in both those states?

Victoria (Legislative Counsel)

Mr. Chairman, Representative Mathy, they are both relatively new. They both had delayed effective dates to give companies time to kind of develop infrastructure to be able to comply with their requirements. So they are relatively new. I think that's definitely something to keep on the radar if they do make changes or if they identify any issues that may require some amendments to this legislation.

Mike Nathe

Was there a lot of pushback from industry on this?

Victoria (Legislative Counsel)

Mr. Chairman, I do believe there was pushback. It would require them to do additional things to comply with the law. So with that, I would assume there is pushback.

Jonathan Warrey

Thank you. Senator Castaneda.

Jose Castaneda

Thank you, Mr. Chairman. I have a question roughly page 3 around line, I don't know, 5 through 7, maybe a little further down, uh, in discussions there about, uh, reporting, because obviously we're not in the child portion yet. I think for children, obviously, you know, you start having issues, they should be reported to parents immediately. Uh, that's part 2 of my question here, but part 1, reporting. I think that we can all agree that if, if someone has true suicidal ideations Reporting to the appropriate authorities is the right thing to do. But my worry with this is what's going to happen when the chatbot, the AI, decides it's somebody they're chatting with has suicidal ideations, they don't, and now this person's in some sort of world of who knows what kind of hurt because now the police have been called, they've been checked, whatever. You know, this transcends to social aspects as well. And I'm a big believer that in the court of public opinion, You're always guilty and hardly ever proven innocent. So I worry about just the reporting portion from a personal privacy perspective, because AI is not always going to get it right. So I don't know if there's any thoughts around the table about this, or any issues have come up in other states. And when, then when it applies to, to children, are the reports going to go to the parents so the parents can do something first before it starts? going out to anyone else? Because, you know, going back to parental rights, are these things going to strip parental rights away and then go straight to anyone else other than parents? So, any— has any of this been brought up?

Victoria (Legislative Counsel)

Mr. Chairman, if I may, I just want to be clear. There is no requirement to report to law enforcement as part of this. There are reporting requirements in, I think, the The design here is to collect data, how many people are reporting to a chatbot that they are expressing, you know, mental health, behavioral health concerns, not necessarily tied to the user. The only requirement tied to the user would be the referral to crisis resource responses. So there's no report to law enforcement for an individual user required under this legislation.

Jonathan Warrey

Go ahead.

Jose Castaneda

And this might be, you know, obviously this is new territory for all of us. I mean, across the country, across the world. But within that same vein then, have any other states put in certain rules in place of where privacy begins and ends for the individual of how much of what you're taking is being reported? I think that something's gotten out of hand. Has been with, you know, without using specific names, other social media websites that unless you go in deep, deep into the settings, all kinds of stuff about you is being sent back to them. So, as we have an opportunity now, I really do feel like we are getting ahead of this as we probably should have maybe 15 years ago with social media. But, you know, that boat has since sailed. And now we're just playing catch-up. What have other states done to find that balance between privacy of the individual and what chatbots can take? And whether— I didn't imply necessarily reporting to law enforcement, but just taking that data and now making it themselves. Any thoughts on that?

Victoria (Legislative Counsel)

Mr. Chairman, Senator Castaneda, my expertise kind of in this area begins and ends with the children portion, but I can definitely do some research to find out what kind of privacy measures have been taken in other states and get back to you to see if, you know, there's a General Privacy Act or something of that nature that could be looked at by the committee.

Jonathan Warrey

And I've got a question, Victoria. If later In the day, if we do vote to advance this bill, a committee bill, and we find something in the subsequent time where we wanted to add language into it, is that an amendment just through the standard legislative process, or can we amend it before we submit it?

Victoria (Legislative Counsel)

Mr. Chairman, the form of the bill as it would leave this committee today would be what is referred to legislative management. Following that, it would be subject to the general legislative process, so any amendments could be proposed All right, thank you.

Jonathan Warrey

Maybe an option, Senator Castaneda, if we get more information.

Jose Castaneda

Thank you, Mr. Chairman. I appreciate it.

Jonathan Warrey

All right, any other discussion on that bill? All right, we will move to comments from interested persons. Just by a show of hands, who's interested in providing comments? Okay. We'll call you up, ma'am, if you'd like to go first. Welcome.

Signe Snortland (Public; candidate, Senate District 33)

Is that better? There we go. Chair Warrey and members of the committee, my name is Signe Snortland. I'm a candidate for State Senate from District 33 in Oliver, Mercer, McLean, and Morton Counties. District 33 is ground zero for hyperscale data center development in North Dakota, with data centers under construction or planned for Underwood, Stanton, Hanover, and Mandan. I appreciate this opportunity to speak to you at your last committee meeting when the rubber meets the road. This is when you tackle how to fix the problems with AI and data center development that you've been discussing during your previous committee meetings. These are my suggestions for data center legislation. First, slow the roll by rescinding sale and use taxes for data center servers, routers, hardware, and computer software. These sales tax breaks have cost North Dakotans $94.5 million so far. It's time to quit giving our tax money to billion-dollar companies and protect our communities from undue development pressure. Second, the 2025 legislature defeated House Bill 1579, which would have required a PSC certificate of public convenience and necessity prior to construction. By converting it into a non-binding interim study and delaying enforceable action until 2027. In the 2027 legislative session, I suggest you mandate some simultaneous state-level socioeconomic review and local county zoning approval for large AI factories and their power plants. This gives PSC state oversight authority while preserving local authority on siting and regulation. Third, enact a statutory ban on non-disclosure agreements that strictly prohibits state, county, and local officials from entering into non-disclosure agreements regarding project locations, utility requirements, power usage, and proposed tax incentives. Fourth, given indications the data center financing bubble is about to burst and technology advancements may render hyperscale data centers obsolete, Require developers to post robust surety bonds prior to construction to ensure the decommissioning and land reclamation. Further, legally authorize local community benefit agreements for local workforce and community investment. Finally, enact tax— ratepayer protection. Again, cost shifting that prohibits electric and gas utilities from passing capital generation or transmission costs of hyperscale data centers onto residential, agricultural, or small business taxpayers. Big business tech developers must fund 100% of their own infrastructure needs. Thank you for serving on this committee and spending time investigating these issues that are so important to the people of District 33.

Jonathan Warrey

Thank you, ma'am. All right, who else would like to speak?

Greg Kessel Sample Client (companion chatbot developer)

Mr. Chairman, members of the committee, my name is Greg Kessel. I'm the executive director of the Children's Advocacy Centers of North Dakota. I just wanted to provide a few comments on the bill. I mean, first thing, we're very supportive of this. Really like this draft. It's something since we first heard about these AI companion chatbots and the dangers, it's been a topic that we really wanted to see some legislation like this. So, Really appreciate this. And these 5 comments I'm gonna make, I hope they won't impede any sort of endorsement by the committee. So this could be something that we discuss more during legislative session if you're interested in these considerations. So the first one on page 2, lines 1 and 2, this is where it says sort of an exception. Is AIs where they're structured as a standalone consumer electronic device that functions as a speaker and voice command interface or voice-activated virtual assistant. They're sort of a type of toy that kids have. I put a couple examples in my testimony here, bondoo.com, miko.ai. There's these— there's toys marketed for kids that have an AI component to them. That build this companionship. Now, and they're all voice activated. They simulate companionship. So are these classified? They're likely classified as voice command interfaces or voice activated assistants. And there's really no evidence showing that the manipulative engagement tactics, that's all that list on page 2, lines 3 through 17. So these devices, these toys, are not necessarily showing these manipulative tactics. However, there has been some concerning AI prompt responses by these devices to children. And I linked to a story there. This is a public interest research group that has done some research on this. And the Today Show did a story. It is worth your 5 minutes' attention to watch this video on what some of the responses are that these AI toys are providing. So would that be an exception? I don't know the answer to that, but I just bring that up for your consideration. Consideration number 2, this is on page 3, lines 21 and 22, where it talks about the operator shall use reasonable methods, including the nature of user prompts, to determine whether a user is a minor. That's that age-gating feature. The user prompts, what does that mean? I think that's the clarity here that we need. The pop-up box that says, I'm 18, and you just click that, is that a user prompt? It likely is enough to work here. But I rarely will highlight what California is doing, but they were one of the first states to have this AI companion chatbot restrictions. And so they've had some experience with its implementation, but they— Yeah. They add a few more details to this user prompt, and they'll hold platforms liable if they ignore clear age indicators like metadata, a link to school email, profile indicators, payment methods. So they add a few more examples in their legislation. And that approach, just adding a few more clarifying rather than just user prompt, It's not as demanding as that third-party age verification model, so it's sort of a good middle ground there. A third consideration, this is page 3, lines 25 and 26. This is the operator needs to issue a clear and conspicuous notification. This is the every-hour notification. Some states— well, let me just say— We all know about banner blindness, where you see a pop-up ad and you just click right off it. You're not even looking at it or thinking about it. So does this every-hour continuous thing just become a banner blindness to kids? There could be other options here, such as an always-on-screen disclosure. And some states have offered both as an option, not just the pop-up every hour, but also maybe a disclosure that's always on the screen saying this is an AI-generated persona. So I just offer that as another possibility there. Consideration number 4, this is page 4, lines 3 through 7.

An advocacy group urged tighter age-gating, always-on AI disclosures, parental notifications with chat log preservation, and civil remedies, signaling amendments that could add cost and liability for the client.

Greg Kessel

This is related to the privacy and account management tools for a minor's parent or guardian.

Greg Kessel

So when operators detect a potential threat of self-harm for a minor, parental accounts This is an idea that these parental accounts could either receive a notification that the, the minor is engaging in this type of conversation. So, and, or the other thing here is that it could preserve the chat logs for situations such as evidence collection that parents, guardians may need in a civil suit. So it, it could be something where we specify here that The account management tool allows parents a little bit more control and being able to see what their children are doing on this, possibly saving the chat logs or getting a notification when the child is engaging in the self-harm conversations. Consideration number 5, this is Representative Schneider who brought this up, the civil rights of action. It just might be another enforcement mechanism so it doesn't— the onus doesn't completely fall on the Attorney General and the criminal enforcement. 2 thoughts here. One is there have been children who have taken their lives from after engaging in deep emotional bonds with these AI companion chatbots. And so having a civil right of way for these families would be good. Thank you. The other thing, this just came to mind as one of these AI companion companies, one of the largest ones, Character.ai, they just went bankrupt a few days ago. And now what are they going to do with those servers? Many people have formed deep emotional bonds with these AI companions. To my knowledge, the company is still functioning. But what happens if they just shut down those servers? You have a lot of people. We're going to need some sort of mental health help. Is there a civil remedy for these families as these young people and even adults are trying to find mental health help to cover for essentially a friend that has died because the company has shut down the server? So that's just another consideration thinking about civil remedies. So I'll stand for any questions then.

Jonathan Warrey

Thank you, Mr. Kozlowski. I have a question. Appreciate the work you do. Dr. Hagerott.

Dr. Hagerott (Citizen member)

Yes, thank you, Mr. Koslowski. I appreciate you bringing up the character AI because it could be the canary in the coal mine. I was reading about that as well. So this is a company that purposely built characters to connect and interact with people, children, whatnot, collected a lot of data. And this gets to the liability thing we'll talk about later. I believe one of the issues is all the data that has been uploaded, vacuumed up with all these people. And if it goes to a bankruptcy hearing, let's face it, they sell it to the lowest bidder, which could be overseas. It could be, you know, for that matter, the Chinese could buy all this stuff. Have you read any more about how are they going to protect all the data in this bankrupt company?

Greg Kessel

Mr. Chairman, Mr. Hagerott, I have not looked or thought much about that. The data center or the data Privacy is a whole other area. As I was looking at those toys, the first thing I mentioned, that was a major concern with those is Bondoo. They had 50,000 chat logs that were just available to the public with a Gmail account. And so all these young kids having these conversations with chatbots, the AI toy essentially was open to anyone and everyone. They've since locked that down. But yeah, there's certainly data privacy concerns along any Representative Nathe.

Mike Nathe

Thank you, Mr. Chairman. Thanks, Greg, for this. This is fantastic, by the way. A couple questions. I really like your 4th point here on page 4, lines 3 through 7, about notifying the parent or guardians and then having a chat log so they could take a look at that. Is this being used in any other states?

Greg Kessel

Mr. Chairman, Representative Nathie, I did not look at any other states necessarily and what they're doing here. So I don't know if another state has done this. Victoria may know. She's done a lot of work. This is just a great idea you came up with one day? I don't know.

Mike Nathe

Okay. Mr. Chair, I really like this one, number 4. There's a couple other good ideas in here too, but thanks for that, Greg.

Jonathan Warrey

Yeah. And Greg, I have a question. As you look at your peers with other advocacy centers, has, to your knowledge, has anybody passed something like this legislation and had some impact yet?

Greg Kessel

Mr. Chairman, in other states, I don't know how many other states have passed it. It's hard to say because it's so new.

Jonathan Warrey

Sure.

Greg Kessel

But I can certainly ask around. And I should have said this up front, but it— so that's the importance. It's on 2 One is the— we work with child sexual abuse victims. So it is that normalizing that adult-child sexual abuse relationship, which a lot of these AI companions do. And then the other part that we offer mental health services to these vulnerable kids. And that is— they are going to these things to ask for mental health advice. We know that. So we just want to make sure there are protections for them as well. But that is a good question.

Jonathan Warrey

Yeah. Thanks again for line-iteming this in a short amount of time and bringing that to our attention. Very helpful. Any other questions from the committee? All right. Thank you, Greg. Would anybody else like to speak for our public comments section? Anyone else? All right. With that, we will close the comments from interested persons. And I think let's move into our next bill.

Aaron (Legislative Counsel)

Thank you, Mr. Chairman, committee members. Looking at LC Draft 27.0359.1000, this is a bill relating to the regulation of artificial intelligence for consumer protection and the creation of a regulatory sandbox program and the Artificial Intelligence Council. It was based off Texas Responsible Artificial Intelligence Governance Act, which regulates how AI systems may be developed and deployed. The first part of the bill goes over the— would create a new chapter to Title 51 for consumer protection. It goes over definitions and then it goes to applicability on page 2. This chapter applies to a person or entity that promotes, advertises, or conducts business in the state, produces a product or service used by residents of the state, or develops develops or deploys an artificial system, intelligence system, in this state. The bill further explains that artificial intelligence systems that interact with consumers shall disclose before or at the time of interaction that the consumer is interacting with the AI. A disclosure must be made regardless of whether it was obvious that you are interacting with AI. This section, this next section goes on to the prohibitions. It says that a person may not develop or deploy AI in a manner that intentionally aims to incite or encourage a person to commit self-harm, harm another person, or engage in criminal activity. Additionally, a person may not develop or deploy an artificial intelligence system for the purpose of identifying a specific individual using biometric data or the targeted gathering of images or other media from the internet. Without the individual's consent. The enforcement— the Attorney General will have exclusive authority to enforce this chapter. The Attorney General is enforcing this chapter with all the powers provided in this chapter and in Chapter 5115. This chapter further explains the notice of violation, the ability to cure, as well as the civil penalty. The next section is Adding additional definitions requiring the information technology department to provide an assessment of the agency's use of artificial intelligence and the impact of the agency's ability to achieve its goals and objectives. The next section is the creation— excuse me, let's see what page I'm on now. It is the creation of the regulatory sandbox program. Which enables a person to obtain legal protection. That is on page 11. It obtains— it enables a person to obtain legal protection and limited access to the market in this state to test innovative artificial intelligence systems for a limited amount of time without obtaining a license, registration, or other regulatory authorization. This program must promote the safe and innovative use of of artificial intelligence systems across various sectors, including healthcare, finance, education, and public services. The bill goes on to explain how a person can participate in the sandbox program and how Information Technology Department will oversee the operations. It further goes on to how a program participant shall provide quarterly reports to the department, and the department shall submit a report to legislative management by July 1st, of each year, which shall include recommendations on changes to laws for future legislative consideration. Finally, the bill creates an artificial intelligence council, which is on page 14. This council is to ensure artificial intelligence development is ethical and in the public's best interests. This— the Information Technology Department shall provide staffing and provide administrative services to council. This section goes on to discuss the membership requirements, which would be 7 citizen members, as well as identifying the duties, existing laws and regulations that impede innovations in the development of artificial intelligence systems, and recommend appropriate reforms and prohibitions on the council. And that is the end of my presentation, so I stand for any questions. Thank you.

Jonathan Warrey

Questions from the committee? Senator Myrdal.

Janne Myrdal

Oh, thank you, Mr. Chairman. Relating to page 14, when you just talked about the composition of 7 members, I noticed that 3 members appointed by the Governor, 2 members appointed by the Lieutenant Governor. I know in Texas those are elected separately. So that's probably why that was included in the Texas portion of this, I would think that that doesn't necessarily need to be the same in North Dakota where they are elected as a team. So just wanna note that that might wanna be changed.

Jonathan Warrey

And I'll just say for the committee's sake, this is a framework and it's documented. It is all subject to the legislative process. But yeah, that's noted. And as well as what it's called, where it reports, how it gets its funding, I think all of that Representative Nathan.

Mike Nathe

Thank you, Mr. Chair. Senator Myrdal addressed my concerns. Yeah, A and B, you're giving 5 appointments to the executive branch and just a couple to the legislative branch. So I think if this bill were to go forward, at least this should be redone. We should figure out exactly how we want to do something like this. But I couldn't— I don't support either A, B, or C. But we can come up with some way to figure out a compromise. Legitimate way to do it.

Jonathan Warrey

Yeah. And Representative Nathie, we'll get you on the record. Do you encourage no senators involved? I withdraw the question. I plead the Fifth. Other questions? Dr. Hagerott.

Dr. Hagerott (Citizen member)

Yeah, Mr. Chairman, I just want to thank your staff and you for including this provision in here.

Dr. Hagerott (Citizen member)

As many people know, I've been thinking and working on AI for a long time, back When my previous job, we did this thing called Envision 2030. And back in 2019, I know we were talking with Ross Gostroly about the emergence of AI and whatnot. And now it's here. But in April and May, I traveled around the state and met with 23 people, business people, legislators. Should we have an AI council like the Petroleum Council? And we have— if you don't know, we have an Unmanned Autonomy Council. As well, a UAS council, which I think we're both on. 23 of 23 people said absolutely. And I've had some of your colleagues call me going, are they going to do that? And so, I just— I really commend you on this. Again, what exactly how, but some group that starts to build expertise, maybe members of this committee, because this is such a difficult concept and takes so much study that you can't expect every legislator to become an expert on this or every business person. So, I just want to commend you on that. Thank you. I commend you to include that in whatever you go forward with.

Jonathan Warrey

Thanks for the comments, Dr. Hagerott. And what I would envision is you've got this AI council that will either— this committee stays as a standing committee as a potential option, or you've got an AI council that has similar duties that are ongoing and reading and reacting to what's happening in the marketplace, other states with legislation. It also encompasses some of the things that our late colleague, Representative Christie, envisioned with an AI council. There's a component that could be added around, you know, the sandboxes to provide, hey, you can't take this to market yet because you don't know if it's steerable and controllable. And so, Dr. Haig-Rott, you're going to advance this potential bill or this concept in our next discussion. But okay, you have up to 3 years to develop your AI before you take it to the marketplace. There could also be a spot for university involvement to say, how are you going to commercialize the intellectual property? And maybe Phase 1 doesn't include that, but it could include some of the work where AI is being commercialized. And in our university systems, our R1s are taking advantage of that. So maybe iteration 1 doesn't have that, but it could be a spot, just like the UAS Council's, you know, adapted over the years and the Petroleum Council has. This could be something that has both protection of our citizens but also commercialization opportunities as well. But we'll— again, that's just the initial framework. I think note that all the sections are subject to our legislative process, but at least this establishes a foundation. Thank you. Representative Schneider.

Mary Schneider Sample Client (companion chatbot developer)

Thank you, Mr. Chairman. I have concerns on page 3 on the enforcement authority that that is unnecessarily restrictive in that in number 1 under enforcement authority toward the bottom of the page that the Attorney General has exclusive authority to enforce the chapter subject to that one exception. On state agencies, and that number 2 would prevent a private right of action for a violation of the chapter or any other law. And the reasons for those restrictions probably are not demanding enough to have it be that limited and restrictive. There are a lot of reasons I think it's important to have it more broadly based than that. So I'd like to see a little bit of work on that section.

A member proposed removing exclusive Attorney General enforcement and allowing private lawsuits, which would greatly widen the client's litigation exposure under the AI consumer protection bill.

Jonathan Warrey

Is there anything specific you'd like, Representative Schneider, given your legal knowledge?

Mary Schneider

Well, I would take out the word in the first sentence, number 1, that the Attorney General has exclusive authority. I'd eliminate exclusive. We want him, of course, to have the authority to enforce And then I would either eliminate number 2 or have the wording changed to this chapter does not preclude a private right of action.

Jonathan Warrey

Do you mind explaining what the elimination of that, or that does not preclude, what does that mean in actuality?

Mary Schneider

That the language as it's stated now does not allow for anyone besides Besides the Attorney General and then the state agency exception there to bring a right of action, and we may want others to be able to bring bring an action if there if there is a compelling need and the Attorney General is not either the the most available or proper party to bring it. This is if this we believe is a good law, we might want others to be able to bring actions, either parents or other private individuals who've been injured by this. And maybe we'll never have that party, but this would exclude anyone else from being able to use this law besides the Attorney General to enforce it. And I'm not seeing a reason why that would be essential to the Any comments or questions from the committee on that?

Jonathan Warrey

Senator Myrdal.

Janne Myrdal

Well, Mr. Chairman, all due respect, I understand the concern, but the Attorney General is— he has exclusive authority in their job to enforce the law of North Dakota. So I think it's just a normal language to say exclusive. The larger conversation, whether there should be civil Action or not, I think it's a large, long-term conversation during session because there's there's a lot of laws that you can have civil action. Other laws you can't, and why Texas chose to put this in would be an interesting thought to find out why why it was added so strongly. So I think I and I appreciate the concern, but I think the discussion on private action will have to take place if this bill. gets forwarded to session because that's a very, very in-depth judiciary discussion on a lot of issues. And so it's deeper than just this particular bill.

Jonathan Warrey

Okay. All right, both comments are noted. Any other questions, comments on the bill? Dr. Hagerott?

Dr. Hagerott (Citizen member)

I'm sorry, I had another one to say. Agreeing with a couple of people about the composition of the committee, I do believe we would want business people on there, building on your point about innovation, because both the UAS Council, Petroleum Council does include businesses because we do want to create wealth. And so just a note for future discussion. Yep, that's noted. Yeah.

Jonathan Warrey

Obviously, you know, this is a 16-page bill being sprung onto your desk. We do get that. But it— this combined with the previous bill seemed to encompass quite a few of the concepts we discussed. More broadly. So those are the 2 that Legislative Counsel had time to prepare. There's been a— so what we will do— any other comments or questions on this? Oh, Representative Novak. Yep, go ahead.

Anna Novak

Thank you, Mr. Chairman. I just kind of wanted to make a comment on your idea for a proposed AI council. I really do like that idea.

Anna Novak

North Dakota, obviously, we're a citizen legislature, which means that we meet 80 days every other year.

Anna Novak

And in this age, this digital age, and technology is constantly shifting, I think it's important to have something with continuity and something that can work on issues relating to AI, which is constantly changing during the interim. Obviously, this could be a—

Anna Novak

you mentioned a committee during the regular session and following through the interim. But I think it's important to have something that's standing that allows some flexibility.

Anna Novak (North Dakota Legislature)

And so your idea for an AI council, I just want it to be noted that I do support that idea.

Jonathan Warrey

All right. Thank you. And it could be enforced to also filter data center siting too. It could be part of their duties, which, which again, as we continue to learn, we're going to adjust to. We don't want to curb innovation, but we don't want to trample the rights of our citizens either. So I think, you know, potentially it could be an extension of what this committee's work is. Any other questions or comments? All right. What we'll do is we will Table that till this afternoon to determine if we'll vote on that as a committee and determine if we want any amendments to those. With that, I would like Dr. Hagerott to advance just for discussion your steerable and controllable draft that you've got. And I think it probably makes sense to distribute that to all the committee members. Say again?

Dr. Hagerott (Citizen member)

You mean the concept paper that—

Jonathan Warrey

Correct, correct. I can talk to it briefly here. Yeah, let's do this. We'll take just a couple, let's take a 5-minute recess and we'll distribute that document to everybody. This will just be for discussion if we wanna see if a committee member would like to help work with you to see if it's something we'd like to introduce for the session. So we'll get that distributed. You were working up to the bell on this, so that's why it hasn't been distributed to everybody, but yep. Yep, we appreciate that. So we'll take a 5-minute recess. We'll get this distributed, and we'll begin discussion on it. Thanks. Ready, Robert? All right, committee, we will call our meeting back to order. And what we have passed out is, again, one of our charges is to address steerable and controllable AI and what that means to us as a state. And Dr. Hagerott, This is an area of expertise for him. I am wondering— well, Aaron has distributed this to everybody. It's on your desk here. Dr. Haighratte, would you give an overview of the AI Safety and Compute Infrastructure Protection Act that you've been working on? We'd appreciate that. We'll have dialogue on that. Thank you, Mr.

Dr. Hagerott (Citizen member)

Chair. So, and I want to thank Senator Hogue sitting over there, reaching out to me in the summer, in June, And then the people I spoke with in April and May going around.

Jonathan Warrey

And one of the things we talked about is the protection of North Dakota, our infrastructure, our hospitals, our banks, our children's data, our universities' data, our ag data on animal research, right? Where is this data? Where is it being posted? How secure is it? And a lot of it at that time was theoretical, like, well, is there really a danger to this stuff?

Dr. Hagerott (Citizen member)

Well, since that time, the President of the United States has come out with called NSPM-11, National Security Presidential Memorandum 11. And the top 2 priorities are steerable and controllable AI.

Jonathan Warrey

Top priorities.

Dr. Hagerott (Citizen member)

Now, why did he do that?

Jonathan Warrey

This past summer, a couple of high-profile events occurred that resulted in emergency meetings of the Federal Reserve, the National Security Council.

Dr. Hagerott (Citizen member)

And that was a thing called Mythos out of Claude Anthropic got out of control. And then most recently, a thing called Hugging Face, which in Unfortunately, the name Hugging Face confuses people, but OpenAI was doing a test of some of its bots in a sandbox, and they made the terrible mistake of connecting the sandbox to the internet. And I was in Washington, D.C., which prompted me writing this up, where both Mr. Bannon, Steve Bannon spoke, and a liberal senator, and they both agreed liability control are now going to among the top of the priorities. And, and so having a little bit of business background and understanding, you know, when, when liability starts flashing around, it was nice to have the first speaker at the podium talking about Character AI that went bankrupt. You can imagine people are going to go for their assets and liability and who's responsible. That this, this proposed concept paper, which could become a bill, is both one, protect our infrastructure, our banks, our hospitals from Reckless models. I'll get into some definitions. But what's fairly novel is also North Dakota doing something to protect our infrastructure.

Jonathan Warrey

And I know there's this tension between the data center people and our infrastructure. Well, I don't know if people knew, but our Public Service Commission used to be called the Railroad Control Board. I don't know if you knew that. It was railroad control because who are these railroads? What are they doing? Eventually, it became the railroad infrastructure was an asset we valued and cared about.

Dr. Hagerott (Citizen member)

And people People wanted railroads to their homes, right? They wanted railroad bridges. And I think that's what's going to come eventually. North Dakota is going to realize this is as much our critical infrastructure as it is these entrepreneurs who are bringing it here.

Jonathan Warrey

And so it has a second provision that we would want to be a place that makes it clear that just because someone has a—

Dr. Hagerott (Citizen member)

what's called a frontier model that's causing damage, hacking money, banks, and whatnot, The data centers themselves are just machines that host data and pump water and have cooling and routers and switches and GPUs, graphical processing.

Jonathan Warrey

But they themselves are not taking the action. And Representative Nathy just talked about China recently was accused of a lot of the cars they want to sell. There's nothing wrong with the transmissions, the wheel bearings, but the actual brain in the car may be controlled by the Chinese army.

Dr. Hagerott (Citizen member)

Think of a tanker truck crossing a bridge that the Chinese take control and they now control the truck. So, so this distinguishes between the structure that we see building out here, which may bring wealth and tax revenue and jobs, and frontier models that literally could recklessly damage things. So just some definitions of the front— frontier models are those models that are at the cutting edge of Human controllability. We're not talking about the data centers that have been in Bismarck or the Bitcoin one that are low-level, human-controlled. There's no doubt humans are in control of this stuff. But we're talking about the models that are being experimented in Silicon Valley that are now showing signs of exceeding human control. And we know they can cause damage. I spend my time in D.C. quite a bit doing some military work. And the major hospital system 7 miles from my other place has been locked up from a cyber hacking incident. They're on the road to bankruptcy. They can only take as many patients as a nurse can sit in the room all day and all night. And they are still struggling to get this out-of-control thing under control. And certain reports are that 70% of all ransomware is now driven by AI models. So this is not theoretical anymore. The president's come out with his NSPM 11, and also had meetings with all the big leaders just a few weeks ago that they agree they've got to do something to assert more control.

Dr. Hagerott (Citizen member)

Second part is to protect our compute infrastructure.

Dr. Hagerott (Citizen member)

And as we all know, when you look at, like, the opioid lawsuits, they were trying to sue anybody and everybody, right? That the distinction between, you know, you don't sue the transmission makers and the ball bearing makers, you go for who had control of the thing that caused the So, there's some definitions. Dangerous frontier models, you know, many of them could be very low level, right? But there could be some that really push the limits, and those are the ones we'd be worried about.

Jonathan Warrey

Give some examples: autonomous goal formation, self-modification. And then the key thing is, can we shut them down and contain them, right?

Dr. Hagerott (Citizen member)

And none other than Jensen Huang, who Runs NVIDIA, has now come out saying— and it's amazing, this guy has gone from 3 months ago saying every one of my employees will have 100 AI agents and they'll monitor them. I don't know about you all, but I would have difficulty monitoring 100 of my employees, right? I mean, and I remember reading and going, are you— essentially, you're no longer in control. He's now reversed that position and said we need high-level engineering to assure control. So he's having a major shift himself. Yeah. So that's one of the key things is control, containment, shutdown. So that's— I think there's a consensus building around that, even from the President of the United States. And so you can see that there'd be some penalties, you know, the legislature would have to decide what they would be. But you'd have to have real teeth, right, if you're going to have people running these systems causing damage. Yeah. And so that's not too unique. What is unique here, and I haven't found any states have done this yet, is beginning to make a distinction between now our compute infrastructure. And again, there's people here who have probably been in bankruptcy business school classes, like how does that work? But, you know, they go after all the assets, right? I mean, and I wouldn't be surprised if this Character AI, they go after all the data people uploaded. That has value, right? They They've got millions and millions of users, all their data, and there'll be a bidding war to get that data, right? Yeah. Well, you can imagine— and this is where I did reach out to TechND to ask if they could have a meeting with some of their people and ask them their opinion on this. Is this helpful? Is it helpful to have at least a statement and some level of legislation that distinguishes that if North Dakota has compute facilities that are somehow connected with some catastrophic event, The distinction of we're not writing the code here, we're hosting it. We have limited knowledge of what they're doing. We have, you know, good governance. We try to know, but in the end, we are not culpable for what may happen. And so that's the second part of that. And then there's some things about empowering the governor. You know, I was talking to one of my friends who lives in a southern state about this. He goes, well, why would the governor need to be involved. I mean, so they shut this stuff down. So the utilities are down for a few days. I said, if our utilities are down for 3 days when it's -30, you are destroying much of North Dakota's homes and infrastructure. So when you live in these rigorous climates, infrastructure becomes a high-level, governor-level issue. And I think also when you think about our banks and hospitals, so, so the governor would have powers under here to defend our digital infrastructure, in ways that we probably never anticipated. And then we get to some positive things here that we should consider having an AI policy development center, this AI council that advises you. And because this is not going to go away.

Dr. Hagerott (Citizen member)

They had a speaker named Stuart Russell. You can Google him. He's one of the leading thinkers on AI.

Dr. Hagerott (Citizen member)

He took the podium after Mr.

Jonathan Warrey

Bannon stepped down.

Dr. Hagerott (Citizen member)

And he said, there's now a high probability— he said 60% chance of catastrophic damage of these systems getting out of control. But he said, do you understand this is never going to end? And I'm a cybersecurity guy. And I remember years ago, 15 years ago, I was in a meeting with an iRobot executive. And he said, once we've solved the cyber problem, we'll be making a lot more And I raised my hand and said, when have we ever solved the crime problem, right? We can shut the jails down now because we— this is never going to end. And that's what he said. And I think that's why the importance of having this committee stay on or a center is that this is going to be what our kids will live with, is how do we live on this planet with intelligent entities that can carry values and norms? And so not only affecting kids, but our infrastructure. And Were you going to give me a chance to talk about the kid legislation too, or did you not want to? After this is done? Okay, I'll stop. Yeah, we'll stay with this topic. And then also some tax credits for AI research. You know, we want positive, make this place a more livable place, just like we regulated railroads or regulated steamboats. You know, it's just going to be part of this.

Dr. Hagerott (Citizen member)

But we also want the business voices because we don't want to be shutting down innovation and wealth creation. It's a balancing So that's kind of the concept paper.

Dr. Hagerott (Citizen member)

It isn't in form of legislation, but I just wanted to get that out there.

Jonathan Warrey (North Dakota Legislature)

Questions or comments from the committee? And what we'd envision here is somebody helping to shepherd this work because a lot of your research was done throughout the time of this committee and working with other peers, which we appreciate. It isn't in a format where we would have it ready out of the committee. But Senator Myrdal.

Janne Myrdal

Well, thank you, Mr. Chairman. I think it's excellent information and we should definitely go forward and draft something and be a leader in the nation on these issues. I think it's also important, I appreciate the presenter separating sort of the business part and the growth of data centers from the AI concerns. Because I think, because we have looked into both on this committee, I think there's a tendency to lump them in together. And they're actually very separate issues when it comes to law writing or lawmaking. I think an AI center committee would be completely separate from a committee or a whatever that discusses the siting for data centers. Those are water and electricity people and backhoes, if you will. It's a little different than AI. So I appreciate that. But I appreciate this work. It's definitely a foundation for going forward.

Jonathan Warrey (North Dakota Legislature)

Senator Castaneda, did you want to say anything? If not, I'd love to get your input. But yeah, so what I wanted our committee, because it was a charge from legislative management, is Develop a framework for steerable, controllable. This is the start of it, but Senator Castaneda, if you'd be willing to work with Dr. Hagerott and the peers on this as a legislator to potentially advance the bill idea for the next session, I think that would be appreciated by the committee. Would you accept that assignment?

Jose Castaneda

Thank you, Mr. Chairman. Yeah, I would most definitely accept. I think this is, like Senator Myrdal pointed out, not only important work, but Truly, I think it's good that North Dakota can be the leader for this kind of legislation and hopefully have some really good legislation out of this next session and then have other states begin to mimic it. And that really helps sometimes push it through. I'd rather see, personally see a lot of states have this type of legislation rather than being pushed top-down from the federal government. So, I think it's an amazing place to start, and I would definitely love to work with Mr. Hegarud on this issue. Thank you.

Jonathan Warrey (North Dakota Legislature)

Yeah, and I think also, you know, you look at Dakota Carrier Network, these infrastructure providers, they're cooperative-owned. Or you look at our banking infrastructure, we got the most community banks per capita. Those also being protected or insulated from some of the bad actors out there, I think, is important. And so that includes that as well. But we had Dr. Hagerott and then Representative— go ahead, Representative Nathie, and then we'll come back to Dr. Hagerott.

Mike Nathe

Thank you, Mr. Chairman. And when Senator Castaneda looks at this, again, the makeup of this committee, obviously legislators, but I think it's important to have experts such as Dr. Hagerott. I think we've seen his expertise be a huge benefit to this committee along with Tony Clark's So whoever makes up whatever committees, I think it's vitally important that we have people such as Dr.

Mike Nathe

Hagerott and Mr. Clark on committees like that to bring that expertise. We could talk about industry.

Mike Nathe

I think that's another discussion. But I just want to, just as a suggestion, let's keep these experts because we have real experts in this field right here in North Dakota that we can utilize and I don't want us to forget that. Yep.

Jonathan Warrey (North Dakota Legislature)

That's noted. Good points. Dr. Hagerott.

Dr. Hagerott (Citizen member)

I was wondering if I could also use this since I'm sitting next to a former ballistic missile bomber pilot.

Jonathan Warrey (North Dakota Legislature)

It's something I think is worth considering is also this committee, you know, we are in an AI arms race and North Dakota has some of the most important installations ever that we also consider Some element of that.

Dr. Hagerott (Citizen member)

And we're talking about indemnifying and supporting the data center industry, that if there's a place that's going to welcome defense compute and all the things that go with it, I think we might want to think about that in part of this scope. Because, you know, this is going to be everything I'm reading, you know, I don't know if you know, but the Secretary of War just established Autonomous War Command. I mean, this is science fiction stuff, a whole command on robots and AI. Yeah. And we have the US Council and all the rest. So I just want to at least make a note about that, that I think there could be something we could say also by leading the nation, because most states are attacking the government right now on this stuff. And it might be an area of growth. And I think General Bussiere kind of encouraged us back in Minot about that too recently to our delegates.

Jonathan Warrey (North Dakota Legislature)

He did. He did. And maybe— is there anybody else from the committee that would also like to work on this with Senator Castaneda? Not all at once. Okay, Senator Myrdal. It's certainly, yeah, from a co-sponsoring, but yeah, I think our August 13th meeting up at Minot State incorporated some of this, and we had Anthony Aguilar, I think, that was helping with that. So, you know, continuing to get some input from those folks would be valuable, I believe, in advance. So Senator Castaneda, thank you for stepping up. We appreciate that. And so committee, we won't have a work product to consider. But we do have some time before lunch, and Dr. Haggott, why don't you— you had done some work