“Building an AI-Ready America: Safer Workplaces Through Smarter Technology”

[Gavel sounds.] Okay. There are a lot of committees going on today and I know we have a service for Congressman LaMalfa as well, so we're going to try to kick things off here. This Subcommittee on Workforce Protections will come to order. I notice that a quorum is present. Without objection, the chair is authorized to call a recess at any time. Today's hearing will focus on AI and other advanced technologies which are improving efforts to keep America's workers safe and healthy. AI-powered safety tools are moving safety management from a reactive incident-based approach to a preventive data-driven model. Wearable sensors on personal protective equipment can monitor hazards like heat stress or exposure to hazardous materials and alert workers before conditions become dangerous. Predictive analytics can focus where accidents are most likely to occur, catching hazards before those hazards lead to accidents rather than afterwards. Understanding how these technologies work will help Congress better understand how they can be used to address workplace safety. The impact could be immense. Fewer injuries mean healthier workers, increased worker recruitment and retention, lower cost and stronger business operations, all while ensuring America's workers receive the highest standard of safety on the job. We also know where there is new technology there are important considerations to go with it. Whether it is AI today or another groundbreaking technology that has emerged in the past, employers should take necessary steps to understand best practices and proper implementation. These tools can be invaluable for augmenting worker safety, but there must be space for human oversight and employers should be wary of delegating ultimate responsibility for worker safety to AI. Moving forward, the adoption of these technologies raises important questions. How can the effectiveness of these tools be validated? How can all parties involved, including employers, employees, and unions, understand the technology's role as part of safety practices? What safeguards are necessary to protect worker privacy while still allowing meaningful data collection and analysis? These questions directly affect trust, adoption, and long-term success. As we've heard throughout this committee's series of AI hearings, effective policy must strike a careful balance between considering these questions without stifling innovation and compromising our nation's technological competitiveness. In the context of AI systems designed to protect workers on job sites, there is no margin for error and one oversight can be the difference between life and death. Ultimately, AI and advanced technologies offer the opportunity to make workplaces safer than ever before. By balancing protection with progress, technology can improve productivity and actively protect people across industries. I look forward to hearing from today's witnesses to understand the real-world applications and impact of these technologies. With that, I yield to the ranking member for her opening statement.

Thank you, Mr. Chairman, and thank you to our witnesses for being here today. Automation is not a distant theory. It is already embedded in hiring, scheduling, surveillance, and safety systems. Right now, employers are making decisions that will shape working conditions for millions of people, often without sufficient transparency, worker input, or guardrails. Automation is not new. However, there is a relatively new way of talking about it with the advent of artificial intelligence or AI. This term is sometimes used to tell a story about automation as both a future that we cannot escape and a race that we must win at all costs. But underneath this story is a very small number of powerful companies with access to massive amounts of data and computing resources. And the lessons we are supposed to learn is almost always the same: that government must stay out of the way and keep its hands off AI. But we do need government to act. From invasive surveillance systems to dangerous work speed quotas, the workplace risks posed by these automated technologies are real and immediate. And under the Trump administration, we have seen an eagerness to grant big tech corporations unprecedented access to Americans' data while rolling back guidance intended to protect workers and promote workplace safety. Weakening those protections does not drive innovation. It shifts risk onto workers and their families, making our communities and our markets worse off in the long run. At the same time, automation and advanced technologies can help make workplaces safer when developed responsibly. A fully funded National Institute for Occupational Safety and Health could lead the way in ensuring these tools are used to protect workers rather than harm them. Strong safety standards can usually foster innovation. One well-known example is OSHA's standard on cotton dust. As a result of that standard, not only did fewer textile workers develop brown lung disease, but the industry also became more productive and efficient. State governments also play a critical role. About half of our states receive federal funding to enforce OSHA standards and their state plan agencies can set standards for hazards that OSHA has not addressed. State health and safety agencies can be leaders in developing important safeguards against technology-related hazards, especially since OSHA's rulemaking process can take anywhere from five to 12 years. It is also not too late for Congress to do its job and legislate. There is already introduced bills that I urge our committee to consider, such as Representative Norcross's Warehouse Workers Protection Act and my bill with Representative Jayapal, the Empowering App-Based Workers Act. We must take an active role in shaping how AI is integrated into American workplaces. AI should be a tool that improves workers' lives by making jobs safer and more sustainable. That means ensuring worker safety is prioritized from the start instead of being treated as an afterthought once the harm has already occurred. Finally, as we continue discussing AI, we must look beyond the workplace and consider the broader impact on communities. The data centers that power AI and other technologies consume enormous amounts of energy, which can strain aging infrastructure, drive up electricity bills, and have adverse effects on the environment. Our constituents deserve greater transparency and accountability about resource use, community benefit agreements that ensure residents share in economic gains, and a real seat at the table as AI development expands. But the truth is none of this works if we continue to hollow out our federal agencies responsible for protecting workers and communities. My Republican colleagues like to talk about government efficiency but then go about it in the most chaotic ways. Recklessly firing safety inspectors, slashing capacity across our enforcement agencies only makes workers less safe and businesses less certain about the rules they need to follow. If AI is going to reshape our economy in the coming years, then we need to make sure that workers and communities truly benefit from and not just bear the cost of this technology boom. Because when we unleash technology without guardrails, workers will always pay the price. Thank you, Mr. Chairman, and I yield back.

I thank the gentlelady. Pursuant to committee rule 8(c), all members who wish to insert written statements into the record may do so by submitting them to the committee clerk electronically in Microsoft Word format by 5:00 PM today, 14 days after this, excuse me, 14 days after this hearing. And without objection, the hearing record will remain open for 14 days to allow such statements and other extraneous material noted during the hearing to be submitted for the official hearing record. And now to the introduction of witnesses. Our first witness is Mr. Yoann Land, the Senior Vice President of Product and Engineering, Safety and AI at Samsara in San Francisco, California. Welcome. Our second witness is Mr. Eric Hoplin, the President and CEO of the National Association of Wholesaler-Distributors here in Washington, D.C. Welcome. Our third witness is Mr. Douglas Parker, a senior advisor at the National Employment Law Project in New York City. Welcome this morning. Our final witness, Mr. Jeff Buczkiewicz, the President and CEO of the Mason Contractors Association of America in Algonquin, Illinois. Welcome today. We want to thank all of our witnesses for being here today and we look forward to each of your testimonies. Pursuant to committee rules, I will ask that each of you limit your oral testimony to three-minute summary of the written testimony which you have provided. As committee members have many questions for you, the clock will count down from three minutes. Pursuant to committee rule 8(d), the committee practice, however, we will not cut off your testimony until you reach the five-minute mark. I would also like to remind the witnesses to be aware of their responsibility to provide accurate information to the subcommittee and with that, I'll recognize our first witness today. Mr. Land, you're recognized for your testimony.

Thank you so much. Chairman Walberg, Ranking Member Omar, members of the subcommittee, thank you so much for the opportunity to testify here today. We appreciate the committee's focus on workplace safety and the opportunity to share how proven, deployable AI technologies are already improving conditions for millions of frontline workers and how thoughtful policy can help accelerate those benefits. While the public conversation around artificial intelligence has focused largely on generative text or office productivity, the most consequential impact of AI is happening far away from the desks. It is happening in physical operations, on the road, on job sites, or on airport tarmacs. And these are the jobs that power the economy and move our goods. They are jobs where risk is constant and the cost of failure is not pecuniary. It's injury or loss of life. At Samsara, we build connected AI-powered technologies for physical operations. And our customers span transportation, logistics, construction, utilities, and public sector agencies. In sectors that represent more than 40 percent of global GDP and millions of Americans workers performing some of the most dangerous jobs in the economy. And we work with some of the world's most complex logistical, transportation, and infrastructure companies and public sector agencies that keep this country running. And we have seen firsthand that in these environments, AI is not an abstract concept. It's a task-based, situational, and preventive tool. And our platform combines sensors and AI-powered video to turn real-time data into actionable safety insights. Detecting distracted driving, alerting ground crews to unsafe conditions on the tarmac, and protecting lone workers to ensure they are not invisible when something goes wrong. And this means fewer crashes, fewer injuries, higher retention, and safer workplaces for millions of frontline workers. For instance, we work with the city of Allentown and Bethlehem Public Works in Pennsylvania to ensure city services are delivered efficiently while keeping drivers safe. But technology alone is not enough. The greatest safety gains occur when detection is paired with skills development. And instead of hours of generic training, AI also enables targeted coaching based on actual events. Workers receive timely, personalized feedback that fits into their day and addresses the specific behavior that matters the most. Consider for a second commercial transportation. Vehicles can weigh 80,000 pounds and operate at highway speeds. Nearly 42,000 Americans die on U.S. roads each year. And in a recent survey that we did of 1,500 professional drivers, 79 percent of them reported experiencing a close call due to distraction in the past year. And AI-enabled safety systems directly address these risks. So across Samsara customers, fleets using AI safety technology see an aggregated 37 percent reduction in crashes within six months. A 96 percent reduction in mobile phone usage while driving and a 69 percent decrease in harsh driving events. And these are measurable operational outcomes. And importantly, these are workplaces, even if they're not always thought of it that way. The same principles apply beyond the road. And Ranking Member Omar, at the Minneapolis-Saint Paul International Airport, our asset tracking and AI cameras give operations teams the visibility needed to coordinate responses and keep runway operations and crews safe. And elsewhere across the country, in construction and field services, AI-enabled wearables and fall detection ensure that if a lone worker faces danger, help can be dispatched immediately. Now, technology expectations in the workforce are also changing. And in a survey of leaders overseeing five and a half million workers, 85 percent said that a modern tech-forward approach makes frontline careers more attractive. Safety technology also improve retention. And after deploying Samsara across thousands of vehicles, DHL, a Samsara customer, saw a 60 percent drop in harsh driving events, but also a 50 percent increase in driver retention. That is, when workers feel protected, they stay. So AI is not the future promise. It is improving workplace safety today where the risk is the highest. And with thoughtful policy, we can expand these proven tools to protect more workers across America and we look forward to working with Congress to do so. So I want to thank you so much and look forward to your questions.

Thank you. Now Mr. Hoplin, we recognize you for your testimony.

Thank you, Mr. Chairman, Ranking Member Omar, and members of the committee. My name is Eric Hoplin, I'm the CEO of the National Association of Wholesaler-Distributors, we call it NAW, and we represent the entire wholesale distribution industry in America. It's about 35,000 companies and they're moving products and goods through the supply chain every single day. All told, we account for about one-third of America's GDP and we employ more than six million Americans. And I have the privilege of every week, I'm in a different distribution center or a warehouse and they come in all shapes and sizes. But really what differentiates these distribution centers are the products and the goods that move through them. So for example, you might be in one that specializes in large building materials. You might see pipes that are the size of this room. You might see huge sheets of metal. You might see pallets full of shingles, heavy shingles, and everything in between. Other distribution centers have smaller building materials: tiny fasteners, screws, bolts, nails, everything everything you might might need to fasten something together. Other distribution centers, they move medical equipment, some move food, some move grass and fertilizer and plants, you've got furniture, everything that you can imagine that moves through the supply chain. And the reason why I'm describing it a little bit is to understand that no warehouse and no distribution center is alike because the weight of the things that are moved, the size of the things that are moved, or the different mix of products that are moved mean that each warehouse or distribution center operates in a different way. I will say though, on the other hand, there are two commonalities really that cross every distribution center in America. And the first commonality is they're all run by people. I'd say that the workforce that is powering America's distribution industry is bar none the best workforce in the world. And the second thing is every distribution center has a culture of safety. Every time I walk into a distribution center, I receive a safety briefing, I receive some safety material before I walk in, and you can tell within two or three steps inside a distribution center what the cultural safety norms are. Everybody is following them, everybody is reinforcing them. And so this has been something that we've been doing for a long time is having these established safety protocols. So as new technologies are coming online, artificial intelligence, one of the things we're very interested in as part of our safety culture is how can we harness these tools to enhance the safety of America's warehouses and distribution centers. Now, it's important to note that we're not the developers of this technology, we are the deployers of this technology. And we're finding that there's four main areas of AI that are helping to improve safety in America's work in America's warehouses. The first is something called disembodied AI. It's AI that is scanning the entire environment of the distribution center to maybe spot in real time if there's a safety incident. Something abnormal is happening, they can alert us in real time, or they can track this data over time and maybe make some recommendations for how we might change operations or things inside the distribution center to make them safer. The second is predictive AI, and predictive AI is using data that can track machinery and other tools to see are they nearing the end of their life cycle. Is something different happening with maybe a forklift that its brakes are running low, running out, and we should replace them before they run out and possibly cause a safety incident. The third use of AI is human-centric AI. It's wearable technology. Maybe someone's wearing something on their hard hat, maybe in their vest, or maybe it's a key fob that just gives them a little alert. Maybe it's a little pulse or a little tone that lets them know that that an incident may be approaching that's unsafe, or or maybe it's ergonomic in nature. If you're bending over in such a way or you're picking up something that's too heavy, want to prevent you from or alert you that this might strain you in some way. And so we use it in that way. And then the final is automated assisted AI, which is a retrieval system where we're using robots and AI to bring the goods through the distribution center to stack them and to store them, but to bring them ultimately to the worker. So as as we go through the hearing and we talk about these in in more detail, as you think about legislating or regulating the artificial intelligence, two things to keep in mind. Number one is one size doesn't fit all because every distribution center is different. We think it's really important that we empower the local leaders who are partnering with their workforce to design the best environment that will be safe for for those those employees in the distribution center. And the second thing is too much regulation can be burdensome and expensive and it can increase the barrier to entry. Most distribution companies are small. They don't have a lot of capital to spend on artificial intelligence, and so don't make it harder for them to deploy that technology that can make their distribution center safe. So in conclusion, we would say that all of us know that AI is changing the world in ways that many of us can't even fathom. But if there's one thing we do know, artificial intelligence is helping us to enhance the safety of America's warehouses. Thank you for the opportunity to testify and I look forward to your questions.

Thank you, Mr. Hoplin. Now Mr. Parker, you're recognized for your testimony.

Thank you, Chairman Walberg, Ranking Member Omar, and members of the subcommittee. I appreciate the opportunity to be here today to testify on how artificial intelligence and other emerging technologies are affecting worker health and safety. For the last 15 years, I have closely tracked emerging technology trends in occupational safety and health from the perspective of a regulator when I was assistant secretary at OSHA as well as at Cal/OSHA, as an enforcer of safety and health standard, and as a promoter of employer investment in safety and health. AI now dominates that conversation about new technology. It is a powerful tool with real potential to reduce injuries and illness. But if AI is not grounded in fundamental safety principles and an ethical framework, it can introduce significant physical and psychological hazards into the workplace. My testimony today will cover three points. First, to be clear, I support new technologies to protect workers. I have witnessed how collaboration between government and the private sector has brought technology into the workplace that saves lives. However, many of the AI tools on the market today focus on modifying worker behavior instead of addressing the root cause of hazards. While these tools can be a helpful part of a safety program, the first question in addressing a workplace hazard is whether the danger can be eliminated, isolated, or controlled through engineering. Skipping these steps and simply training workers to adapt to hazards makes workplaces less safe. And safety-purposed AI should not be used to discipline employees. That undermines safety. Second, new technologies should not introduce new hazards into the workplace. Algorithmic management tools that track worker activity can create physical and psychosocial hazards. Fear of surveillance, loss of privacy, and anxiety about job loss affect mental health. Physical injuries and illness can result from production pressures that cause workers to skip needed rest breaks or work at unsafe speeds. These risks can be reduced through prevention by design. Practical strategies for prevention by design in AI include development of risk assessment tools, auditing standards, and a safety certification process focused on ensuring technology does not harm workers. And employees should have a seat at the table. Transparency and understanding are key. My final point today is that the government must build its capacity to understand and to interact in an AI-enabled workplace. The question is not whether to regulate AI, it is how to prepare agencies to address the impacts of AI so that they can carry out their basic missions in an AI-enabled world. Thank you and I'm happy to answer any questions.

Thank you, Mr. Parker. And now Mr. Buczkiewicz, I recognize you for your testimony.

Thank you, Chairman. Chairman Walberg and Ranking Member Omar and members of the committee, for giving me the opportunity to participate in today's hearing exploring such an important topic: how we can use technology to keep our nation's employees safe and productive. My name is Jeff Buczkiewicz, I'm the President and CEO of the Mason Contractors Association of America. We represent a $41 billion industry that employs about four million employees all from coast to coast. When we got started, we actually are developing AI programs, AI safety programs for our employees. As an industry, we decided we needed to build tools and programs that actually work specifically for the masonry industry. I tasked my team to go innovate and we have spent the past four years building George, a purpose-built AI system for the masonry industry. George is named after our first president of the MCAA many years ago, about 76 years ago, and takes pictures and videos of job sites to immediately tell my members, their foremen and employees, if their wall bracing plans and implementation meet safety regulations, safeguard job sites during the initial phase of the wall being assembled, ensure in real time employee PPE compliance so that remedies are immediately implemented instead of focusing on a what-if or what happened after an accident has occurred. And be a virtual storehouse for best practices and safety regulations that can immediately be recalled at a moment's notice. Over the next five years, estimates are showing that the construction industry will lose over 40 percent of our workforce to retirements. Construction is hard working. We must embrace systems that not only keep our employees going home safe every night, but they can also use it to increase productivity, keep the wear and tear of construction work to a minimum, allowing longer careers, and allows them to safely complete their jobs more efficiently, saving time and money. Technology is allowing young men and women who might otherwise not think or be able to join the masonry industry to join the masonry industry. Technology is ensuring our employees are working efficiently, productively, and safely. Technology is a tool that can immediately change the landscape in how we train our employees, whether new to the industry or new to a best practice. Technology analysis has the potential to completely change the way we learn about job sites and what new equipment or practices are actually being utilized or not and why. The positives are endless. My one ask of you here today would be to partner with industry in a bipartisan manner to ensure that technology use is encouraged in a responsible manner. Not to be used to punish, not to be used to replace workers, not to be used just to keep to be keeping and staying relevant, but used to increase employee safety and protection, used to prolong careers and productivity, used to learn how employees go about their jobs and how they can conduct them in the safest way possible. We are sitting at the forefront of a truly groundbreaking moment, but for it to take off, we need partisanship between industry, or partnership between industry and government. Thank you, I look forward to your questions.

...balancing that with constructive improvements. That's a key thing to bring buy-in.

Okay. Thank you. Mr. Hoplin, your written testimony mentioned several types of AI-powered technology such as disembodied AI, predictive analytics, and human-centered AI. How do these technologies work together to improve workforce safety in your industry?

Thank you for the question, Mr. Chairman. They work in unison. They work together because they address different parts of potential workplace safety risks. And so what it ultimately creates is a layered approach to safety. So for example, with the disembodied AI, it's monitoring the environment. As it's monitoring the environment in real time, it can spot if there's an abnormality and if we need to alert somebody that there's an immediate safety risk. With the predictive AI, it's monitoring, it's tracking data. And so the data is going to predict when something might happen so we can get ahead of that and prevent that from happening. The human-centered AI is the wearables that employees are using so in that moment, they can find out if they're not using an exoskeleton, if something's going to be too heavy or if they're moving in a way that ergonomically can harm them. And so they're addressing different parts of the risk and together they create a layered approach that makes ultimately the environment safer.

Can the employee override some of these if they choose to for whatever reason?

I believe that they do have those options to do that. But as Mr. Parker pointed out, those systems will be designed working with the workforce of how should we implement them in the distribution center. And so the employees will have a say in before they deploy that technology.

Thank you, Mr. Chairman. I yield back.

Thank you. Next we'll go to Ranking Member Omar from Minnesota.

Thank you, Mr. Chairman. In recent years, we have seen troubling examples of AI-enabled surveillance harming workers. Whole Foods reportedly used an internal heat map that ranked stores by unionization risk. And many warehouse workers have criticized Amazon's use of invasive monitoring software, putting their health, safety, and privacy at risk. A 2023 study found that 53 percent of always pressured Amazon workers were injured compared to 28 percent of those rarely pressured. Mr. Parker, you noted in your testimony that these automated management tools can create fear of surveillance and job loss. Can you explain how AI-enabled monitoring could lead to real health and safety risk for workers?

Thank you, Ranking Member Omar. Yeah, I'd be happy to. Let me first begin with the mental health implications of some of the more aggressive kind of surveillance methods that you are referring to. I think that anybody who has had experience with working on the job understands that even though they're in an employer setting, there is some level of privacy that intuitively they expect. And they certainly do not expect constant surveillance. And so these kinds of methods can have a kind of damaging and draining effect on a worker. They can create stress and anxiety. If it's directly related to performance, then they can certainly have greater anxiety about job loss because they fear that these levels of performance measures are putting their job at risk. I, when I was at OSHA, I spoke to an Amazon worker who showed me a little laminated card that showed each of the steps in the packaging process that they were to use when they were packing a package and sending it on for delivery from their workstation. And it showed a 37-second period of time in which they were required to complete their job. Now, Amazon denied that this was their policy and that they denied that there was a quota. But it almost doesn't matter because if the employee believes that they're under that kind of pressure based on the tools that are being used, they're going to work faster, they're going to be at greater risk of injury because you're introducing strains and the possibility of working to failure biomechanically when a worker is under those kinds of pressure.

And Mr. Parker, you have the unique perspective as someone who ran both federal OSHA and state OSHA agency. Can you start by briefly reminding the committee about what the OSHA state plan program is?

Yes, the state plan program is in the OSHA Act. It permits a state that wishes to have its own OSHA program to apply and if it meets certain conditions, then federal OSHA will delegate its authority to that state and the state agency becomes responsible for enforcing and regulating health and safety laws on the job in that state.

And what would happen if states were prevented from regulating any workplace safety hazards related to AI and how long could workers have to wait before any protections were in place?

Well, I think that it's kind of impossible to answer because it's inevitable that the things that states and federal government currently regulate are going to be impacted by AI. And they can't meaningfully do their jobs and carry out their mission without understanding AI and being able to do regulation in ways that are necessarily going to implicate these tools that have been integrated into their work. I mean, if OSHA can regulate a conveyor system but that conveyor system is algorithmically managed by AI, they can't separate those two out in their thinking about the system.

Thank you. I yield back, Mr. Chairman.

Thank you. Next we'll go to Mr. Grothman from Wisconsin.

Thank you. Mr. Hoplin, a common concern about AI is that it could make important decisions without human input. In your written testimony, you emphasized the importance of human judgment in the decision-making process. How should human judgment be included in workplace safety monitoring and decision-making?

Thank you for the question. And it's very important to have a human in the loop through the entire process. And as we're designing our safety systems using AI, we ensure that that's the case. It starts with the architecture. And it starts with the common safety practices that we've established over many years. And so we'll work with the workforce and with the leaders there to talk about what sort of AI technology can we use to enhance those already existing procedures and processes. And so it's designed with humans involved from the beginning. And then as the AI is deployed and it's giving us data or insights or ideas, the humans are ultimately making the decisions about what we're going to do with that data information. So having humans involved in the entire loop is important in how we deploy it in distribution centers.

Can you give me an example of a bad thing that would happen if humans weren't in the loop?

Well, a lot of bad things could happen if humans weren't in a loop. But we're not deploying agentic AI that's making its own decisions. The AI in our distribution centers can't override the humans. And so we have that loop system where humans are in charge.

Yeah. You can't give me an example? Well, okay.

We don't deploy it that way.

Okay. Mr. Land, one of the biggest concerns I have over adopting this innovative technology is the upfront capital needed, leaving small and mid-sized businesses behind the curve. Will these technologies become more affordable and available over time, you think?

Yeah. Thank you so much for the question. Yeah, I absolutely think so. There's a set of costs driving this. One is of course the hardware cost that is likely to come down over time. There's cloud cost, compute to deploy the AI. That is rapidly decreasing right now, probably about the factor of 10 per year. So yeah, this will absolutely come down and become more affordable.

Okay. What must happen for technology to become affordable? Is there other than the fact that this is what always happens with new technology? We all know that. But they want me to ask what must happen for technology to become more affordable and available.

I think a set of things. Companies like mine and together with policymakers, we need to enable the market and companies and employees to really embrace this. I think that's a key thing to ensure that this has the impact that we're already seeing. Broad deployment and adoption is a key element.

I'm going to ask the question I asked Mr. Hoplin of you. Can you think of bad things that happen if there's too much AI and not enough human?

So thank you so much for that question. I agree with Mr. Hoplin that human in the loop is absolutely instrumental and that's also how we do it. And without that, yes, bad things could potentially happen. Now it's worth noting that in our case, we're not doing self-driving cars or anything like that. We're purely analyzing data, informing employers and employees about that to help them improve and stay safer. We're not taking any actions.

Could you give me an example of a bad thing that has, even you're aware of outside of your own company, that has happened because we have a lack of human supervision over the AI?

Yeah. I don't want to speak to what has happened in other places. But in my company, I have no examples of that because we are so rigorous in our human in the loop and our monitoring of what the AI does.

Okay. We'll give you the two guys on the right. Do you have any examples, scare stories that you can give us that either your or somebody else's company did not provide human supervision here and a bad thing happened and you thought, geez, how did we let this happen?

I can't think of anything nor would we have ever advocated for AI to be implemented without any type of human oversight. I think it's critical that you have that. Otherwise you will have errors and you will have problems and you'll see it more often than not.

No examples? Not even some... go ahead.

I can give you an example of a technology causing harm. It was an investigation when I was at OSHA. A worker in Ohio was working in a manufacturing plant. He was working alongside a robotics device. There'd been some maintenance to the robotics device and its base had been pivoted. It had been trained to work within a certain zone. And unbeknownst to the worker, because the device had been moved and there'd been a failure to reprogram the device, it engaged, it crushed him against a wall and he was killed. So that's an example of humans being in charge but also being complacent about ensuring that they are actually actively engaged in engineering out failure.

Thank you. And I guess I'm out of time.

Thank you. Next we'll go to Mr. Casar from Texas.

Thank you. Mr. Land, it says on your corporation's website that you have a live in-cab monitor feed of truck drivers and that employers who use your product can access months of footage. And I just want to understand specifically what that means. So I have two yes or no questions. For the entire time an employee is driving a truck, if their company is using your technology, those drivers are recorded from head to knee level. Is that correct?

Thank you for the question. That is not correct. Depends on installation of the camera. It depends on the settings of the customer, both the employer and the driver, on how the privacy is handled inside of the vehicle. And privacy is designed into the system to protect for those types of situations.

To protect for what kinds of situations? I'm just asking, are you recording, you're recording the driver?

When the camera is turned on, then yes, we're recording the driver in accordance to the settings that the driver and the employer have set forth.

Got it. And so do any companies set it so that they're being recorded for the entire time they drive?

It is possible to do that. I would say that it's very uncommon. It's cost-prohibitive to do that.

Got it. But a corporation could monitor them the entire time they drive. And then the second question. According to a Canadian arbitrator, those recordings don't just capture the drive itself but in at least some cases could capture 15 minutes after the vehicle has stopped. Is that correct?

Also here, this depends on the setting. But we do enable recording a little bit after the vehicle has come to stop because often safety is particularly important as the worker leaves the vehicle and does work in the job site. So we do enable behind settings that the employer and driver agree.

Right. So the employer could record continuously for the full drive and there are instances, as you've mentioned, of recording afterwards. So I just want to ask about a couple of potential examples. So again, I understand that this, I want AI to help us be more safe. But what I want to understand is in an example of a driver driving for hours, misses a call from his doctor, stops his truck, and immediately returns that call, could an employer's settings record that conversation with the doctor? If they're in the cab, they park the car, they call their doctor.

Depending on the settings...

It sounds like the answer is that it is very much possible, right?

The aim of the technology is really to improve safety in that...

I understand the aim. I'm asking about a very quick question which is can you say here that they would record the doctor's call if the settings are that the recording happens even when the truck is turned off?

Yeah, I understand the question. None of our customers are using the technology in that way. I've never heard of any...

Are you saying right now under oath to us that you know that there have never been a recording of a doctor's phone call in any of these trucks?

I said something different. I said I am not aware of that and I'm not aware of any of our customers using the technology in that way.

But you know that it's possible. But it's certainly possible.

It is theoretically possible, yes.

And also if somebody called their union rep or their spouse or their child. That's also possible.

I have never heard or had any indication of anyone using the technology in that way.

Here's my concern. I know that this is not what you say your software is for. But isn't it true that an employer who wanted to find out who's trying to organize a union or wanted to find out who has a health condition or who's about to have a kid, that it would be much easier to do if they had continuous AI-powered monitoring accessible to them? Wouldn't it be easier to do?

I don't see that being the case. You need to remember that these are extremely dangerous jobs and workplaces.

I want AI to make our jobs safer. Let me ask you this question. How many AI-powered continuous monitoring cameras do you have in your office for yourself?

For myself in my office where I work. I am actually not aware. I hope that we have many cameras to keep the workplace safe.

I'm saying do you have AI-powered cameras in your office? I don't.

I believe we do, yes, because I believe we use our own products in our own office. I'm not completely sure but I believe that's the case. And I hope that's the case to keep us safe.

So you hope that you have a camera with AI-powered in your own office at all times.

Absolutely. Yes. You need to understand this is reversing trends when it comes to safety in a more impactful way than we have seen since the seatbelts or the airbags. This is transformative technology.

Well, I will yield back with after one more sentence, Chairman. I think that the vast majority of Americans would not tolerate having, especially folks that are CEOs or higher-powered folks, 24/7 monitoring with AI capabilities. We want AI to be able to keep us safer. But I have not heard in this conversation any clear guardrails to keep it from being invasive. Thank you. I yield back.

Thank you. And next I will turn to myself for five minutes of questioning and would actually like to continue with Mr. Land. So you have talked about how including sensors and AI-powered video analysis in your written testimony can turn real-time data into practical actions to improve safety. Can you please expand on the actual use of the products that you have and how that's improving safety?

Thank you so much. I would love to. This is something that is absolutely transformative as said. So the way it works is that we install cameras in the vehicles. These vehicles record what is happening and locally on the vehicle run models to identify unsafe situations to help the driver. This could be too close following distance or the driver about to fall asleep and we wake them up. And this saves lives on an unparalleled basis. Now then we analyze this data also with larger models in the cloud to spot patterns. Example of that would be spotting continuously drowsiness. And we can actually detect sleep apnea from this. These are transformative situations for these employers to help them stay safe. And that is why they absolutely want the cameras and as I said earlier, refuse to go back in the vehicles without it.

Thank you. Next I'm going to go to Mr. Buczkiewicz. During our series of AI hearings, we have heard about the potential for AI to reduce costs and make products and services more affordable. How can AI when applied to improving workplace safety efforts actually help reduce the costs in that space?

Sure. No, thank you for the question. Yes. It can do it in a variety of different ways. Our wall bracing program itself has two different components to it. One is an internal bracing monitor that actually will look at some criteria that you enter into the program and it will tell you if your wall can be internally braced versus externally braced. A lot of walls do have the steel and grout in them now which makes them not need external bracing. So if you don't need that external bracing, you're going to save a lot of time and a lot of cost by only using the internal bracing. Conversely, if you do need the external bracing, it'll let you know quickly that you need the external bracing and move to the external bracing and you'll enter the same criteria and a few more details and it will tell you exactly where you brace it. So instead of having the field guys have to go through our book on wall bracing standards that has a lot of engineering formulas in it that are not the easiest to understand and try to calculate that formula and find out where the best place to brace their wall is, the program will tell them where to brace the wall. They can go out and brace that wall. They can take a picture of it as soon as they're done with their phone and upload it right away and the program will verify that that wall was braced properly. If that wall happens to fall during a windstorm when nobody's on the property, hopefully, that wall falling over will be very costly. It'll be hundreds of thousands of dollars to replace it. So that's just one example of how it can save costs.

Sure. Well, I appreciate both of you sharing your experiences about how we can improve health outcomes for workers and improve workplace safety. I've seen it in our local community. Our health network is utilizing this in radiology and oncology when they were doing cancer screenings. They were having AI run applications in the background. About 10 to 11 different AI programs were running. And while I was sitting there, they actually detected that somebody had come through for a screening on a kidney or liver issue, actually had a blocked artery in their heart. And they said we need to call this person back immediately. The doctor didn't even actually do that screening himself but it was picked up by AI in a bigger screening application. And he said ultimately that person is probably going to have their life saved because otherwise they would have had a heart attack offsite, may have been rushed here from a rural area, may or may not have been actually been able to be saved. So I think we're seeing great applications, not to dismiss anybody else's concerns about privacy. I think we obviously want to make sure that that is maintained as well. But I think the applications for improving health outcomes and worker safety are really incredible with this AI technology. And so we want to figure out how to responsibly advance that as we continue to move forward in the technology improves. Next I'm going to turn to Mr. Fine from Florida for five minutes of questioning.

Thank you, Mr. Chairman. I'd like to move in a little different direction. We heard a lot of concerns about the idea of AI being used in trucking, I guess, because of some hypothetical doctor's visit. I'd pose it that the Democrats are concerned about AI in trucking for other reasons. I want to talk about where AI might have been useful. Perhaps had we had AI in trucking, an illegal immigrant from Kyrgyzstan who failed to brake and swerved his semi-truck into a van might not have killed four people. I suspect that wasn't the first time he drove and he didn't know what he was doing. Perhaps if we had AI in a truck in October of last year, an illegal immigrant from India would not have killed three people on the road. Perhaps if we had AI in trucks, an illegal immigrant from Serbia in October would not have killed one person in Indiana. And in my own state in August of last year, perhaps had AI been involved, an illegal immigrant from India would not have made an illegal U-turn in the middle of the highway and killed three people. I suspect in all of these instances, it was not the first time they drove inappropriately. Now I understand the Democrats care more about illegal immigrants than they do about Americans. But I would suspect that the use of AI would help with safety. Now I've driven a Tesla for the last 10 years. And I've taken advantage of the driving features that it offers. In fact, now I can drive hours and never touch the steering wheel. And that's largely because of AI technology. I can only imagine if it can be used for me what it could be done for other people. It's got collision warning, lane keeping, driver monitoring. These all help accidents in real time. Even when I want to drive myself, having these things on, it's like a second set of eyes to help make sure that people are safe. Mr. Land, do you believe, you discuss in your testimony AI-enabled coaching and training and you reference reduction in crashes and distracted driving. How do workers respond to personalized training based on monitored events?

Thank you so much for that question. And I very much share the your sentiment around the power of AI in preventing accidents. It can truly be transformative. And AI is core to this. We have around 70 billion minutes of data to analyze. And the pattern spotting through that is enormously hard to do in any other way than AI. And that is what leads to this more than 70 percent reduction in crashes. So it's enormously powerful.

Okay. What lessons have you learned from the employers who've successfully implemented this stuff?

Yeah. So we see drivers in particular embracing this. And it is because it is tailored to help them improve. It is spotting the things that they truly need help with rather than bringing drivers into a classroom watching a mandatory 10-minute video. We tailor it to what they need, personalized, combined with positive reinforcement. So we see drivers vastly on board with this and passionate about it.

So I want to turn my attention from driving to the warehouse issue with Mr. Hoplin. I think some of the testimony you hear from people who've actually never had jobs in the real world, like they pretend there are no cameras. So I worked in the casino business for more than 10 years. And by the way, I guarantee that just about everywhere there were cameras. Those cameras are there both to secure the validity of the games, make sure people don't get ripped off, they're there to look for health and safety, and they're there just for the security and safety of people. And you have armies of individuals who watch these cameras in large part because that was the only way it could be done. I'm sure a lot of things are missed and frankly there's going to be bias brought into it anytime that you use people. An example again of an industry and by the way, I was a CEO and there was a camera in my office. I'd like to ask how Mr. Hoplin, how burdensome regulations would make this more difficult for you to do this. How burdensome regulations make it harder for you to roll these things out in warehouses and other places.

Thank you for the question, Mr. Fine. I think one of the important things to think about is wholesale distributors, 80 percent of the industry are small businesses. They have 20 employees or less. And so they only have so much resources that they can deploy to new technology. So they might pick one or maybe two that they can deploy to enhance the safety of their warehouses because that's important to them as well. But as they're making those decisions, they got a lot of questions to ask. What rules are the government going to implement? Do I have to hire some attorneys and accountants to make sure I'm following all of those rules? If my distribution center is in Minneapolis but Minnesota has a different rule than Wisconsin, I got another one over in Wisconsin. And so as there's a patchwork of regulations, it increases the barrier to entry. And so what it does is it reduces innovation and I think most of those smaller distributors, they're just going to kind of wait before they deploy that.

And I would share your view as I wrap up. AI will be a tool particularly as it scales and it can operate on easier and easier to access computers that will allow small businesses to do many of the things that larger businesses already do today. It's our job to make sure the government doesn't get in the way. With that I yield back.

Thank you. And with that, that concludes our questioning of the panelists today. And so we will now turn to closing remarks. I will recognize the Ranking Member for her closing statement.

First I just want to say we shouldn't insult the intelligence of people on this committee by saying casinos have cameras. We know they're not voice-enabled. And so your privacy still remains if you were to call your doctor. And two, just because you are not a U.S. citizen, it doesn't mean you're undocumented illegally not in this country. But thank you, Mr. Chairman, and thank you to our witnesses for speaking with us today. As we close today, I want to emphasize a simple point. Automation and AI will shape the future of work whether we act or not. The question is whether we shape it intentionally and carefully or allow workers and communities to bear the risk by default. Technology itself is not the problem. The problem is deploying powerful tools without guardrails, transparency, and accountability. When that happens, safety becomes optional, risks are shifted onto workers, and harm appears long before policy making makers respond. History shows that when government steps back and we do not get neutrality, we get rules written by those with the most power. Congress does not need to choose between innovation and worker protection. Strong safety standards have repeatedly driven smarter and more responsible innovation. Clear rules create certainty, reward good actors, and ensure new technology actually improve working conditions instead of making them worse. That is why federal leadership matters. Fully funding NIOSH, strengthening OSHA, and supporting state plan agencies are essential to keeping workers safe as technology evolves. These institutions help ensure that automation reduces injuries, protects privacy, and makes jobs more sustainable. Technological progress should not come at the expense of workers' livelihoods, health, and safety. We must also recognize AI's impact extends beyond the workplace. The rapid expansion of data centers cannot come at the expense of public health, environmental stability, and shared economic prosperity for all. AI should be a tool that works for people, not one that extracts value and burdens workers and communities with costs. Congress has a responsibility to act now before the harm becomes unavoidable and impossible to solve. Thank you and I yield back.

...AI and other advanced technologies are actually protecting workers in the current situation and about the future applications how we can do even better. There is a clear opportunity to create safer workplaces, prevent injuries, and enhance smarter decision-making, and we see that when employers act thoughtfully and collaboratively with workers. We also heard how AI-powered technology is most effective when paired with a strong safety culture, human judgment, and worker trust. These principles should work hand in hand with innovation, especially when lives are on the line. And again, I'd like to thank all of you as we came together to have this hearing because I think there are great opportunities as we move forward. We can do that in a collaborative fashion with industry and with Congress, but at the forefront of all of this is going to be worker safety from this subcommittee. And so again, would appreciate your continued involvement and thank you for being here today as we continue our series on hearings about AI in the workplace and how we can actually continue to make America more prosperous and a better place for all of our workers to grow and raise their families. Thank you again, enjoy the day. With that, I would like to close the hearing. Without objection, there is no further business and the subcommittee stands adjourned. [Gavel sounds.]