H.B. 276 Artificial Intelligence Modifications: transcript

Yes. Thank you, Mr. Chair. Members of the committee, and I will be speaking to the first sub. Um, so this bill is a result of a coordinated group effort conducted in partnership with the Department of Commerce and the Office of AI policy. I have Zach Boyd here. He's the Director of the Office of AI Policy. We worked over the interim, um, on this issue of deep fakes. Um, we worked deliberately to study the AI trends most affecting our children and our citizens. Our goal was to search for reasonable balanced solutions to some of the real and emerging harms associated with deep fake technology, while also working collaboratively with industry stakeholders to ensure and stakeholders to ensure the framework is workable and responsible. So, HB 2 76 does two primary things. First, it addresses the creation of intimate images generated through AI or deep fake services. The bill requires large technology platforms to obtain consent. If an intimate image is created using generative or deep fake tools, we know that non-consensual I intimate imagery can cause devastating and lifelong harm. At the same time, we work closely with industry partners to build in safe harbor provisions and liability protections. So platforms are shielded from frivolous lawsuits when they're acting in good faith and following the law. Second, the bill addresses the growing inability to distinguish what is real from what is digitally altered online. HB 2 76 requires covered platforms to provide users with a way to view metadata and data provenance. In other words, the history of how an image was created or altered. So individuals can better determine whether, uh, they are seeing what they're seeing is authentic or manipulated. In addition, the bill includes a provision directing the Department of Technology Services to work toward disclosing provenance data on government websites. This allows the state to lead by example and helps establish a broader industry standard for transparency around digital content members of the committee. There is an urgent need to protect our citizens from the damage and confus confusion that deep fakes can cause intimate images, especially deep, fake, intimate images can ruin lives, and an inability to trust what we see online can erode public confidence and deteriorate the social fabric of our communities. HB 2 76 is a surgical and targeted approach to addressing some of the most pressing risks associated with deepfake technology, while maintaining balance with innovation and indu industry feasibility. And with that, uh, Zach and I are open to any questions. Do, do you wanna add anything?

No, absolutely. Okay. Nailed it.

He's a, he's a good pro, so he, he has plenty to add.

Thank you, Mr. Chair. And thank you for bringing this. I actually thought this was already illegal, and maybe you could just describe the, where this statute will fit in the field of legislation and related area, if you're okay with that. I, I actually did. I had, I've been under the impression this was already illegal, and anyway, I'm supportive. And then I had one other question, the civil penalty, I'd go even higher than that, but I'll vote for this. Okay. But anyway, maybe touch on how you thought about penalties and Yeah. How those could be, uh, well, yeah, thank you.

Yeah. Yeah. Well, so it's, it's illegal for children and it's also illegal criminally, this creates a private right of action for the platforms. Um, I didn't wanna, I didn't wanna be too heavy handed with the penalties. Um, I'm hoping the industry will be responsible and move in this direction, but this will nudge them that way. Um, but yeah, I mean, you're right. It is serious. It's, it's serious. But, um, we are trying to strike that right balance with industry.

Okay. Yes. I'm sorry. I'm, it's getting late. So you, you're saying that it was criminally, it was already illegal criminally, but now we're providing a private right Of action. Yeah. Is that correct for adults? Is it, it's illegal criminally for adults, is that correct?

I believe so. Yeah. So we were just creating that private right of action for platforms if they aren't obtaining consent for those images to be created. Yeah,

I'm very supportive. I'll just add one comment in support if that's appro, if, yeah. Um, you know, I, I I think we need to go further and just say, AI in generated images are really problematic. They ought to be identified in some clear way. Um, there's such potential for abuse. Um, this is only the most extreme aspect of it. I think we ought to go further eventually. Uh, and I'll, I'll leave it at that. Thank you. I like the bill. Thanks.

Thank you. I have a couple of questions. Uh, could, could you talk about the difference in, uh, intimate versus non-intimate images in the requirements for consent?

I can, let me, um, do you want, you want me to tell you What makes an image? What, what

are No, no, what, what are the requirements in, in terms of getting consent for a non-intimate image? I think we all understand the importance of, uh, getting consent for an intimate image. But let's, uh, use the example of, if I take a picture of the us uh, or the, uh, US Supreme Court, and I put, uh, Representative Owens and Representative Miller, uh, and, and tell AI to put their faces in a picture of the Supreme Court. That's not an intimate in image, but I, if I didn't ask them for permission to do that, what are my responsibilities?

That's a fantastic question. Do you, I don't know the exact answer to that. I mean, the, the first chapter of this bill is really focused on the intimate image question for non-intimate image is we have kind of a business use law that was passed last session. And Senator Colmore is running, uh, an i, uh, rights to your personal identity, um, bill this session that addresses that from a different angle. And I think, and Mikel did that last? Yes, Mike, yeah. So Senator Mikel and Senator did it last session too. So we, we have addressed that and are addressing it, but that's not contemp.

So you, so you believe this bill has no effect if somebody's just creating, let's say, a, a, a funny image to share with their friend and they use their friend's image, that is, that is not covered under this deep fakes bill.

We're only concerned about intimate images, Correct. I mean, the, the second chapter may apply in terms of provenance, but that's, that's not a consent requirement. Yeah. I mean, you would be able to tell if it's real or not by accessing the data provenance. However, there's no limitations on whether it can exist or not.

Uh, may, maybe that's worth explaining. Uh, if, if, if I put their images on a Supreme Court image and I've, I've asked, uh, Gemini to generate that for me. Um, how, how will they, uh, follow the, uh, the provenance rules in this bill?

Zach's gonna say it really well.

No, what, what Gemini and other industry actors are moving towards is just putting a latent meaning, you know, invisible to the human eye signature in the, in the file that cryptographically verifies that an AI created that image. So it wouldn't necessarily have to obtain personal data, it wouldn't restrict anyone from creating the content. It's just a transparency standard.

Thank you. And, and can you speak to, this obviously takes some time for platforms to prepare for this. What, what, what are the timeframes you've got and the expectations of how this would be verified?

Yeah, yeah. We pushed out, um, the date to, uh, January 1st, 2027. And for one, and for one part to 2028, because there are some technical bits that will take a little bit longer.

Uh, chair, I, uh, request leave to do a couple of things. Uh, first thing I'd like to do is, uh, move to adopt the first substitute of House Bill 4 38. That motion is in order to adopt, first substitute, uh, well, 4 38. I think we're on oh 2 76, Correct?

Yes. I withdraw that motion chair. And I appreciate the clarification.

I, uh, ask, uh, I, I hereby move, uh, first substitute of House Bill 2 76.

I'll waive.

Vacated. I will move the amendment.

to the motion to move amendment one? No, I will not particularly. I'll,

Yep. This was just a clerical error, or, and we're just fixing that and making it consistent with definitions and other part of the bill. So it's a friendly amendment.

Well, let's, let's go for the hat trick. I hereby move, uh, that we favorably recommend how's Bill? Uh, 2 76. 2 76. First substitute as amended, uh, to the house floor to their, uh, for Their consideration.

I'll, uh, waive Back to the, to the sponsor.

Thank you Committee. I, I do think this is an important step as we continue to monitor artificial intelligence as it evolves. I don't think this is where it'll stop. We'll continue to watch how this technology, um, develops and fill in the holes where we need to, to protect our children and our citizens. And I urge your support. Thank you.

I Vote. You may. Thank you, Mr. Chair. Um, no man or woman has a right to do through an alliance with others. Anything he or she doesn't have the natural right to do alone. And doing so is a harm to all. So I'm a no

Good morning, Mr. Chair, members of the committee. Um, I have with me director Zach Boyd, uh, the director of the Office of Artificial Intelligence Policy in the state of Utah. He is a wonderful partner, and as many of you may know, we have an AI task force, um, made up of some representatives and Senator Vickers. And we work closely with the Office of AI policy over the interim to look at trends in AI and what we can do to better protect our children and protect our citizens from, um, the emerging harms of ai. Um, as well as looking at how can AI help to improve lives and how can we, um, encourage innovation in a safe and responsible way. Um, so, uh, director Boyd is here to answer any technical questions because he is the technical genius in this group. So, um, so I just wanted to tell you two things that this bill does. So the first thing the bill does is that it requires platforms to obtain consent if you're going to deep fake an intimate image of somebody else and, and put publish a, a deep faked intimate image on a platform. Um, we have worked closely with industry, I think we've talked to almost every, um, large tech company. And, um, we've tried to protect them from frivol frivolous lawsuits and create a safe harbor. Um, for those who are good actors in this space, uh, and those who are good actors and who are adhering to their policies, um, are protected from such frivolous lawsuits. The second thing this bill does is it, um, addresses our growing inability for, uh, of our ability to distinguish what's real and what's not, um, with deep faked images and videos. So what this requires is for platforms to disclose data provenance or metadata. And what that is, is it's the history of an image and how it's been altered. Um, so you would just click on that like a little watermark and you would see the history of that data and you would know this has been altered in such and such a platform or not. And what you're seeing is real that will help you to determine, um, if you can trust an image or not. Uh, finally, this does direct the, um, director of the Department of Technology Services to work towards disclosing provenance data on government websites so that we can lead the way in industry. Um, and so those are the things that it does, and happy to answer any questions. Thank you. That concludes your presentation.

Okay. So you're just talking about images, you're not talking about the counterfeits. Uh, I mean, you listen to radio and you'll hear them have imposters, uh, representing whether it's Donald Trump or, or Arnold Schwartzenegger or something like that. You're not targeting that industry, you're talking about the images itself. Correct?

Correct. This, I mean, this, correct me if I'm wrong, I mean, there is some audio components. Just go ahead. The, the, Yeah, toward the end there is some touching on audio, but the bulk of the bill is about the image and video, Although we have, we have addressed audio in the past. Um, so that is already covered in code.

So how do we enforce this? Or who en enforces it? I mean, you get hit every day. So what's the mechanism? How does the public know who's gonna enforce this?

Yeah, so this is, this creates a private right of action. Um, so the platforms, uh, this would give someone a private right of action against the platforms if they were not obtaining consent or not disclosing data provenance. Um, then, and we also have an enforcement mechanism through the Department of Commerce.

No, that's right. The private right of action is for people who are non consensually deep faked, and then there's consumer protection enforcement for the, uh, labeling stuff.

I understand what we're trying to accomplish here. I'm just having a hard time following it to the point that are we ready for it?

And just, I believe we are. We, we have, so we've started down this, we've created the Office of AI policy. They work in conjunction with the Department of Consumer Protection, and we also work with the Attorney General's office, um, to make sure that this is already to go. So we've kind of built an infrastructure that's ready for it.

Oh, thank you, Mr. Chair. Um, I just wanted to respond to the good Senator Vickers questions. Are we ready for it? And I think not only are we ready for it, we need it, um, because people are out there doing these ne nefarious things, and I mean, I, I wouldn't wanna see it on my kids, my daughter, my loved ones. And we need to start now, uh, to make sure that there are, um, the processes in place for us to go after these bad, bad actors. So, I, I just wanna commend the good sponsor for bringing this forward. Thank you. Thank you, thank you.

So that's why I'm wondering. So thank you. I mean, we do have to address this and it's really hard to anticipate everything, but I know there are many examples out there of, um, people really being caught in these deep fake situations and they're very intimate pictures, et cetera. And I worry about not just kids, but adults, uh, et cetera, and that kind of environment. Um, as you think about this, is this, I mean, on this continuum, where does this put us on the continuum of being able to at least try to protect individuals, uh, from having experiences like this? Is this more about at least giving them private right of action so they can do something about it? Um, you know, what do they do and when, when it happens to them, who do they call and what do they do? Because that's what most people wanna know, right? Um, and they don't jump to private right of action. So any thoughts about that?

Well, it depends. So if, if for example, someone's deep faked, then they can issue a take down request to the platform that has the deepfake on it, so that can at least get it down. If the platform doesn't comply, then they have a private ride if they want to. Also, the generation service, most of the generation services, although this is kind of complicated, are now actually marking which content they created. So some generation services can also be approached because they, you know, produce content they weren't supposed to. But really the take for a person on the front lines, the take down request to the platform is the main line of action.

Okay. So we are kind dependent on the take down request working. Um, and then how are we communicating that to, uh, to individuals?

So, well, so the take down is a federal law. Um, so how that's communicated, I don't really know how to, to tell you. You probably know better than I do, but,

Well, I, I, I don't, and I, but I, but I just think we have, um, this is a good first step. Um, but I worry a lot about our ability to, um, to do things that are harmful, um, to individuals. Um, and when you know it, it's one thing to have political conversations or whatever it is, another to do things that are harmful to individuals, um, and how they are able to, um, be able to respond and respond quickly enough, et cetera. So, yeah, thank you for what you're doing and just keep thinking about what else we do To help. Thank you. Well,

and I appreciate that sentiment. I think the goal in this bill is that, um, those who are good actors want to set an industry standard so they can be trusted and reliable sources of information. And so if they're willing to set that standard, then people know what they can trust and it'll make, and they'll lean towards those types of platforms. This goal is to help the companies who want to be good in this space start going that direction. And it is, it's happening somewhat organically, but this is going to help continue in that direction. Um, I think like anything you, you want to, you wanna eat food that you know you can trust, you want to use a service that you know you can trust. And so the other ones kind of fall off, and this is kind of helping, encouraging that, that free market behavior, um, in a non, in, in a government smartly regulated way.

Good morning, Mr. Chair. Good to see you. Margaret Woolley, Bussey, executive Director of the Utah Department of Commerce. I'm just coming up here to just, you know, just to kind of refresh everyone on the, the creation of the Utah, um, office of Artificial Intelligence policy, which you all created two years ago in statute. And we've stood up and our department, and you, you heard from, uh, Dr. Boyd, one of the things that that office is tasked with doing is identifying places where there may be emerging harms from ai, or places where we may need to look at removing guardrails, uh, because of the nature of the technology. And we really saw these harms of, um, of deep fakes as being a huge problem in our society, in our economy. And, uh, and so, and the reason, of course, that you all can see is that it, it really destroys trust. When people don't know whether images are, um, fake or not, it can destroy trust. Um, and so that's, that's really the motivation here. There's many deep fakes that we see, that we all kind of know are made with ai. They're unrealistic. We, we sort of understand that. But the idea behind this content provenance is that if you don't know, and it matters a lot, maybe it's a political thing, it matters a lot, then you need to be able to know whether or not it has been altered with ai. So this is really all about trust. Uh, one of the reasons I think there is so much anxiety about AI is this issue, and we want to be able to, to make sure that there can be trust in this technology. Uh, and so that's really one of the motivations behind these recommendations, and we so appreciate, um, Representative Defay leadership on this, her passion for this. She actually brought forward the notification piece. We think that's absolutely outstanding and, and we're really grateful for that. So thanks.

chief Privacy and Director of the Office of Data Privacy. Uh, many of you're familiar with our efforts on the privacy side. Utah has been very unique, though one of our foundational efforts has not just been saying your information's private, but part of the future of privacy is nobody can impersonate you with the data that has been breached. So this bill does have some very, uh, important new concepts being introduced specifically around provenance. And that means verifiability of where did the data come from, where did the images come from? And if you are not the source of that, or not the one that has verified that, that you authorized it, then that data would not be able to be associated with you to say this is something that, that you allowed or was created by you. So in my view, this bill is critical to introducing that part of the concept to ensure individuals cannot impersonate you, but you are the one that's in control of your identity or images or f facilities created about you. So I think it's a great bill.

Good morning committee members. My name is May Kennedy. I'm the chief of staff for the Abundance Institute, which is a policy focused nonprofit based here in Utah that works on AI and energy policy. And I'm delighted to be back with many of you here on the hill. Um, the Abundance Institute has a couple of concerns with this bill that we wanted to bring to the committee's attention. We're sorry, um, to the sponsor. We haven't engaged prior to this hearing, but this bill moved faster through the process. And, um, being, not being up here every day, uh, I was more out of the loop than I anticipated. But to start, we agree with the goal of the sponsor. Um, non-consensual, intimate, deep fakes are harmful and deserve strong consequences. Uh, our concern is not with that goal, it's with the structure that this bill uses to accomplish that goal. It creates a state level technical mandate for a AI provides. The internet doesn't stop with Utah Borders. 50 different state standards will fragment compliance and discouraged startups from operating here. Key terms like widely accepted standards, reasonable safeguards and technical feasibility are vague. That means uncertainty, litigation risk, and higher compliance costs. Big incumbents will manage to comply, but smaller builders will avoid Utah, which hurts innovation and competition. This pushes government into regulating the technical architecture and presentation layer of online speech. This raises constitutional and preemption concerns, especially as AI continues to develop the content. Providence provisions risk create a liar's di dividend where fake imagery with faked Providence data is created, is treated as genuine and elevated by DISTRI distributors as well. A better, a better pathway forward is to align with federal law, punish bad actors directly, and support voluntary open standards rather than creating a Utah specific concern. We're happy to work directly with a sponsor, um, regarding our concerns, but appreciate the committee's considerations.

my name is Scott Stornetta and I'm the CEO of Sure. Mark Digital Identity Services a Provo based company. My way of background, I've been involved in the artificial intelligence field for more than 30 years. In fact, mine was the first PhD that Stanford University granted in the field of neural networks. Neural networks being the foundational technology for all AI deep fakes. Our company provides protection for business executives from real time impersonation attacks and the consequences of financial deep fakes. It also protects you as public officials through a generous grant from Zions Bank Corporation and U v's. Institute. Shemar is also a member of the Coalition for Content Providence and authenticity, sometimes referred to as C two PA. I also teach on the intersection of AI, technology and society at Ray Kushner Yeshiva High School. And I'm considered by many to be the co-inventor of the blockchain. I speak in support of the proposed legislation. It is well researched and thoughtful, including its victim remediation framework and its definitional precision. It provides appropriate balance in respecting the rights of victims, while providing reasonable safe harbor protections for legitimate businesses. More broadly, it provides another example of Utah leading the nation for responsible legislation in the area of AI and related fields such as digital identity. Shemar made a conscious decision to locate its headquarters in Utah, in part because of its forward-looking legal framework. I recommend you vote in favor of this bill. Thank you.

Thank you so much. Is everyone able to hear me there? Yes, you're good. Great. Uh, chair Harbor, members of the committee, thank you so much for having me here today, and I apologize for not being able to be there in person. My name is Nathan Trail and I'm the Vice President of Sacred Government Affairs at the Information Technology Industry Council, or I-T-I-I-T-I represents 80 of the world's leading technology companies. Our members are also, uh, the leading AI companies, many of whom operate here in Utah and are deeply committed to developing responsible and innovative AI systems. So to start, and, and, and just to make it very clear, our members fully support the intent with this bill and have been actively engaged in the process to address several outstanding issues with it. They're participating in participating in good faith, and the remaining requested changes are reasonable technical clarifications intended to improve the bill's, workability and its effectiveness. These include aligning definitions with technical standards, ensuring that technical requirements that do not exist today at scale are not mandated before they can be complied with and ironing out safe harbor requirements within the bill. We appreciate the delayed implementation data. Is es especially important that statutory language governing rapidly evolving technology technologies be carefully calibrated to ensure that the state can, can continue to lead on innovation rather than unintentionally constrain it. These proposed revisions are consistent with approaches being considered in other states and are intended to align with emerging federal frameworks. We respectfully request additional time to continue working with stakeholders and the legislatures to ensure that this bill ultimately supports rather than impedes technological innovations in Utah. So with that, we respectfully request more time, and I thank you so much for your time here today.

I'd like to move that and we favorably recommend first

Yes. This is a technical amendment, one that was brought to us by industry. Those who have engaged. Not everyone has, um, we've tried to engage with as many industry partners as possible, but I think this amendment will help alleviate some of the concerns.

Uh, I'd like to move that, uh, we send first substitute of House Bill 2 76 as amended to the Senate floor with a favorable recommendation.

Thank you Mr. Chair. I, I really do support what you're doing here. Um, but a lot of the questions that were brought up leave me concerned with, um, lawsuits and chaos in the AI world. When we do not have tight laws, we actually create problems. So, uh, I will support this now, but I hope that you can work with some of the concerns that were brought up with some of these speakers. So thank you for the work you're doing and I hope you can continue.

Yes. Uh, I do have my concerns too, like my colleague there, um, we had a bill go through your Senate sponsor has a bill, kind of like this. And as I read this bill, I thought we already addressed it, but there are different issues and I see that now. So I have some reservations on whether or not it's time to pull the trigger to, to move this. You know, my preference is that we send it to interim and get it right to start with, but I will support it at this time.

Thank you, Mr. Chair. I would encourage you, Ms. Kennedy brought some good points up. I would encourage you to get with their group and make sure that they're, if possible, their concerns are met. Thank you. And I'll just comment looking through this, I've read it several times and you've got unlawful distribution, you've got safe harbors in here, pleading standards, civil. I think the intent is good, but I still have some reservations on how the bill is crafted itself. The difference between the civil liability and the criminal and, and how that, uh, operates on, uh, some of the local, and especially on the national, international level. So I do have a little bit of concern, again, not with the intent, but with the construction of the bill.

Representative for summation. Yeah, Thank you. And I appreciate the concerns. Um, I just want to assure you that we have worked, so this, we did work. This was an interim, uh, item. We worked through the interim. Um, we have worked extensively with stakeholders. As you notice, we have a heightened pleading standard and a safe harbor. Um, we've tried to do everything but happy to continue working and to make sure that this bill, uh, makes it to the finish line. I think, like you said, we all agree with the intent. Um, we've been very thoughtful and deliberative in working with everyone and happy to continue that. Thank you and I appreciate the support.

I, I just wanna say, I think this bill, um, is pretty narrowly focused on things that for many people would be like the worst case scenario. If this were your grandchild or someone else that you knew and this happened to you, um, this is pretty narrowly focused on things you're not gonna wanna see happen. Do we have it? Perfect? Probably not. But we don't get most things perfect up here the first time, and particularly things that are very early, but to start to bring some protection around it and to begin to work in the right direction. I would also encourage you to continue to have conversations with people, but I wanna protect my grandkids and, and others from this kind of bad actors and, uh, having themselves, uh, through no fault of their own kind of, uh, exposed and very, um, vivid ways. So thank you for doing this.

Yes. This is a DeepFakes bill. Um, and this is just, we are, we're addressing large platforms. We are just making a few modifications so that it's not touching tools that are, um, that are making changes to images. This is more of a generation generative type tool that we're, we're looking at. So we're just making those clarifications in the substitute, I mean in the amendment.

Thank you. As I mentioned, this is a date DeepFakes bill. So this bill, um, was drafted in coordination with the Office of AI policy in the Department of Commerce. We worked over the interim, um, with the legislative task force, uh, to try and find some targeted solutions on how we protect our children, how we protect our citizens from the emerging technology of deep fakes. And so this is, um, from that effort. So the first thing it does is it addresses the creation of intimate images, um, that are created from AI deepfake services, or what some people are saying, notify, notify apps or large platforms that are using generative services, um, to create those types of intimate images. What it does is it requires those large platforms to obtain consent, um, from the person who's being deep faked. And if they are not, then they need to take it down. Um, and then it, we've also just worked with industry to make sure there's a safe harbor for all of the actors in this space who are doing everything they're supposed to be doing, following the law and complying. The second thing it does is it, um, requires the disclosure of provenance data or metadata. And I'll explain what that is right now. We don't know what's real and what's not real when we're looking at images or videos on the internet because of deep fake technology. So what provenance data is, is it's the history of how an image or an or a video is changed. And if you click on that data, then you can know if something is real or not. So what this does is it requires platforms to disclose that provenance data of an image. So now we have a way to know if something is real or not. In addition to that, this also includes a provision that directs the Department of Technology Services to disclosing provenance data on government websites so that we can help lead the way, um, in setting, in setting an example and, and trying to get industry to move towards being transparent and trustworthy and where, so people can know what they're looking at, whether it's true this, this is an important bill, it's important for protecting children, for protecting each other, and helping us, um, gain trust back in the things that we're seeing online. And also to protect, um, from intimate images that can be created that can be severely harmful for, for families, for individuals, and can have lifelong effects. And with that, I'm open to any questions.

The substitute does two things. First, it clarifies that if a generation service does not offer the ability to create counterfeit images at all, regardless of whether consent is obtained, then that service is not subject to the consent requirements or, uh, in order to qualify for the safe harbor. And then second and clarifies the requirements related to provenance data. A platform that allows users to download provenance.data is not required to provide the entire content file or separate version of the content itself.

Thank you. So this bill has been dubbed the Digital Voyeurism Prevention Act. Um, it addresses the rising threat of DeepFakes and the erosion of digital authenticity. The act provides civil and legal recourse for victims of non-consensual AI generated intimate imagery and mandates a technical framework for digital content provenance to identify whether digital media was human captured or AI generated. So it prohibits, uh, AI systems from creating counterfeit, intimate images of identifiable individuals without the individual's expressed consent. It requires large platforms to remove reported non-consensual counterfeit images consistent with the federal Take it Down act. It provides a right of action to an individual whose counterfeit image was distributed. Uh, or if a platform fails to remove the reported image, it mandates, uh, recording devices and generative AI systems to embed metadata to distinguish between human captured AI and synthetic synthetic media. And then finally, it requires digital content or state government websites to contain provenance information, meaning where this originated, if there's a risk that fraudulent or misleading media could cause harm to a Utah resident engaging with the website.

Thank you. The fiscal note has caught up to this now. So this bill deals with deepfake liability, prohibits AI systems from creating counterfeit, intimate images of identifiable individuals. Uh, it requires large platforms to remove those consistent with the federal Take it Down Act. It provides a right of action for counterfeit intimate images distribution. If, uh, if, uh, if reported, if, um, it mandates that recording devices and generative AI systems embed metadata to distinguish between actual, um, photos and synthetic media. And then it requires digital content on state government websites to contain provenance information. If there is a risk that fraudulent or misleading media could cause harm to a Utah resident engaging with the website.

I'll move that under suspension of the rules we pass. Second substitute house bill 2 76. Thank

Motion to Santer for 15, maybe 20 minutes. So Moved. Motion to Saunter for 15 to 20 minutes.

of the president, Madam Marine Clerk Second substitute House bill 2 76. Artificial intelligence modifications by Representative Defay.

Motion to concur with the Senate amendments. Thank

Yes. The amendments do two things. First, it clarifies that if a generation service does not actually offer the ability to create a counterfeit and intimate image, then it would automatically qualify for safe, safe harbor. Secondly, it clarifies Providence related to downloading Providence data. Um, if a platform offers data, it, it originally said they had to be able to download the data with the provenance. This clarifies that they don't have to download the data of the actual material, just the metadata and provenance data.