Committee on Labor - Mar 26, 2026

Everyone, sorry for the delay. We finished floor a little bit late but now we are ready to go. The Minnesota Senate Labor Committee is called to order. I'll note that we do have a quorum and we're ready to begin with our agenda for today. First up we have an overview and confirmation of the members of the Board of Electricity. And we have Trevor Turek here, if you please. Come on up. Chair of the Board of Electricity, welcome to the Senate Labor Committee.

Quickly, thank you very much for that great presentation. Members, do we have any questions or comments or discussion? Okay. Well, thank you. This is really informative. It was really nice to have that overview. Members, you should have in your folders a couple of packets, and one of the packets is certificates of appointment. It's a packet of them and additionally information about the people who I would like to have us move to appoint today and refer that recommendation to the Senate. So if would would people like a moment to be able to look through those packets before we make that motion, or are people comfortable moving forward now?

Yes. We don't have a candidate here.

I have here. Is it Keith Kovard we're looking at? Yes. So maybe what I'll do is I'll read out the names that we have. They are in the packets. So the motion, if we're ready to make it, Senator Pappas, would be that we would move to recommend to the Senate the appointment of Keith Colevard, Sarah Gunmanson, Steve Hybe Hay, Jeff Heimrill, Travis Thull and Desiree Weigel. We would move that they all be confirmed.

Because usually they come and they say a few words.

Don't want that or they're not here, or

Yeah.

Right.

Thank you. Thank you, Chair. Thanks for everybody's time this morning as well. I wasn't as fancy as Trevor here today, but maybe a little old fashioned and I wrote something down. But I was asked as the chair of the High Pressure Piping Board to give a brief overview of what we do and what it pertains to. On the Minnesota High Pressure Piping Board, we oversee three critical facets of mechanical piping systems that are key not only to our state's critical infrastructure, but to the nation's that require highly qualified staff and contractors to carry out the installations. The three primary focuses are the piping systems, our bio processing, steam and other heating media, and ammonia refrigeration systems. Minnesota state law has strict standards to qualify businesses and individuals who design build these systems. We hold the industry to high standards to ensure our key industrial manufacturers are installing their systems with the proper engineering and qualified staff, which inevitably ensures them to operate and a safe manner. We focus on key national standards that are adopted to be the foundation of how we govern our installations. Our board is composed of inspectors, engineers, pipefitters, contractors, one public member and often people from utility companies. Together with our diverse backgrounds, we make rulings with everyone's vast knowledge of the industry as a whole.

Thank you for your time.

Aye. Aye.

Yeah.

Absolutely. Thank you. Chair McEwen, Senators, for the record, I'm Aaron Rosenthal. I'm the research director for Northstar Policy Action. We're a think tank dedicated to improving the lives of working Minnesotans. I'm here in support of Senate files 4576, 4686, and 4689. Some of you may remember I was here last month presenting findings from a report I co authored with Dr. Manjeet Rege of the risk of AI to Minnesota's workforce. I'm not going to repeat that full presentation today, but I do want to just briefly reiterate five key points that I think are pertinent to the legislation today. First, AI capacity and adoption are accelerating rapidly. Today, eight in ten companies use AI. That increased adoption reflects increased capability, and AI is only going to get better. Capacity for AI is currently doubling roughly every seven months. That means if we wait until just next legislative session to take action on this issue, these systems will be four times better than they are today. Second, workers are increasingly anxious and they're asking for government's help. Surveys show that Americans top concern with AI is not misinformation or even privacy. It's the workplace. A simple rule. If 80% of people are telling government to regulate something, you should do it. Third, Minnesota's workforce is particularly vulnerable. Our analysis found that one in three Minnesota workers is in a job with high AI exposure that is the highest share in the Midwest. But even that number understates the issue when it comes to AI being used in hiring decisions or Workplace surveillance. Every Minnesota worker is at risk. Fourth, Minnesota has largely stood still while other states have acted. States across the political spectrum, Republican and Democrat, are already putting guardrails in place around AI in the workplace. We as a state are falling behind. And fifth, in remedying our inaction, Minnesota should focus on areas of greatest concern for workers. What I called in my previous presentation, the three Ds of AI risk in the workplace. Displacement, digital surveillance and decision making on displacement. We're already seeing early signs of job loss and it's likely to get worse. The CEO of Anthropic, the AI company behind Claude, is predicting that half of all entry level white collar jobs will be eliminated within the next one to five years. While we don't know exactly how fast this is going to happen, we know enough to prepare. Workers facing job loss will need time, support and real pathways to retrain on digital surveillance. The solution is simple. Employers should not be collecting data that is unrelated to the job. And workers should know what data is being collected about them with the ability to access and correct it. No one should be managed by an invisible system they don't understand or even realize is operating. And on decision making in the workplace, there must be transparency around the use of AI. And AI should never be the primary decision maker in important employment outcomes such as hiring and firing. Lacking this protection, we risk creating a workplace where decisions are made by a black box with no accountability. While all three of these areas are addressed in today's bills, I would also encourage this committee to continue thinking about one other issue. Boosting worker voice. AI innovation and worker welfare do not inherently conflict. But aligning them requires that workers who are the experts in their jobs have a say in how the technologies are used. Listening to workers will lead us away from a path we followed too many times in past periods of technological disruption, one of wealth concentration and worker exploitation. That is where we're headed. This time. Let's embrace a future of worker satisfaction, increased productivity and economic flourishing. Thank you.

That sounds great, Madam Chair.

Well, thank you, Madam Chair. Members. I will preface the introductions to all the bills by just saying I don't hate AI. In fact, I think if it's done right, it has tremendous promise and it could be used to accomplish great things. And, you know, I think we're on our way to accomplishing some great things. However, consumer facing AI with no consumer or worker protections really is a five alarm fire for our society and it has potentially devastating consequences, especially when it comes to our employment conditions. So Senate file 4689 and 4686 in particular help set standards that allows companies to deploy innovative technology like automated decision systems or electronic monitoring, while also protecting our employees from discrimination, surveillance, and essentially keeping a human at the helm of our workforce. Let's start with 4689. Senate file 4689 puts common sense, appropriate and much needed safeguards on automated decision systems being deployed in the workplace. Automated Decision Systems, or ADSs, are AI tools, essentially machine learning algorithms and data analytics that can assist or increasingly replace humans when making decisions about hiring, performance evaluation, promotions, layoffs, job assignments, discipline or termination. When automated decision systems are useful, while they can be useful in theory, as with all AI, there's a tremendous risk of bias, threat of surveillance, and the threat of hallucinations, which is why common sense safeguards outlined in this bill are so important. This Senate File 4689 would require companies to disclose to workers before deploying an automated decision system. It would prohibit employers from using automated decision systems to obtain or infer private information, including health or reproductive history, disability, religious or political beliefs, immigration status, or veteran status, and prevents employers from using automated decision system alone in employment decisions. This bill also creates due process for workers to access information being input into the automated decision system and to know how that was used, used by the system to make decisions, as well as the right to appeal and have a human review the decision. Moving over to Senate File 4686, this would put basic guardrails in place to protect employees from unnecessary and harmful surveillance. The definitions in Senate File 4686 around artificial intelligence, automated decision system, electronic monitoring tool, employer employment related decision, and so forth are essential. They're the Same actually as 4689. And this bill would allow employers to use electronic monitoring tools to ensure compliance with laws and regulations, protect the health, safety and security of the workplace. But it also makes clear that employers are not allowed to surveil their employees for the purpose of preventing their employees from following the law to obtain private information we talked about before, health status, disability status, religious or political beliefs or make predictions about employees beliefs, intentions, personality or emotional state. It prohibits them from taking action against the worker for exercising their legal rights. It doesn't allow them to monitor employees in what I call universally recognized private areas like bathrooms or locker rooms, pumping or nursing rooms, and religious areas. And most importantly, it prohibits employees from monitoring their employees outside of their workplace or from doing their work duties. This bill, like the one before it, requires pre and post deposits deployment notification for employees and the right for employees to access their own data as well as appeal to a human any decisions that are made from the collected data. And last, there's a really important piece to the electronic monitoring bill, Senate File 4686 that says that data privacy and security is something we still care about and it prevents data from being sold or transferred to a third party unrelated to the analyzing and interpreting of the worker data for safety or security security. And then last we have Senate file 4576. And this regulates employer requirements for when an employee is displaced by AI. The bill requires a 90 day notice if employment affects 10 or more employees or 10% of the employer's workforce, whichever is less. And it requires the Department of Labor and Industry to develop templates for these notices. It entitles employees to transitional employment periods, clarifies that the commissioner or attorney in general can enforce this law and and requires an annual report to the legislature to be submitted on the number of covered employees that met the displacement notice threshold and summary data on the workers and worker classifications affected. And so I'll end with this members, again, I do think AI has tremendous promise. I think it can be used for great things and we want to be sure that technology can be used by our employers. This just puts in some really common sense practical safeguards for our employees. So we know that the field we're operating in is relatively stable and what we would expect from our employers.

Good afternoon, Chair McEwen and members of the Committee. My name is Nico Montag and I work at the Minnesota Department of Revenue. I also serve as a Chief Steward for AFSCME Local 2829. I am here to voice AFSCME support for Senate files for 4576, 4686 and 4689 and I'd like to spend some time telling you about what AI implementation looks like at my job. I'd also like to thank Senator Mayaquaid for authorizing this legislation. The Department of Revenue adopted the use of AI software, Amazon Connect, in September of 2025. If a taxpayer calls MDOR today, they will complete a call not knowing that the conversation was summarized and transcribed by AI. Any worker data that is generated from this call is also captured by Amazon Connect, which includes the full audio recording of the conversation and a full video recording from the worker. The transcripts and all other data of these calls is retained for at least six months. We are also subject to monitoring of all of our keystrokes, what's on our monitors, the quality, brevity and accuracy of all of our conversations with the members of the public, and our timeliness on completing clerical work. When a call ends, we even have to change our online status if we get up to go to the bathroom. Needless to say, we are a very heavily monitored group of workers. MDOR's Quality Assurance team measures the quality of all of our work by randomly selecting calls that they can replay and score us on a variety of criteria. Some of these metrics are very clear, like checking to see if we set our names and that the call caller reached the Department of Revenue. Others are a bit more nebulous, but at least we know that a human is the one evaluating us and that frankly, not every one of our interactions is going to be scrutinized. We might be constantly surveilled, but at least we are being watched by human eyes. But the AI tools that the state has implemented don't just have a sample of a call here or there. It's capable of analyzing every call and even our private moments and compiling these results for our supervisors to review. Amazon Connect can evaluate our vocal intonations and even be used to evaluate expressions. Amazon Connect's materials were told to us upon its first release that they potentially could be used to evaluate the customer interactions and make sure that they could be evaluated for potential positive, medium or negative interactions. So it is our understanding that the state will likely expand this AI software across the enterprise. And I think that before this system expands any further, workers deserve to know whether or not we will be evaluated and potentially disciplined with AI tools. So please support Senate files 4686 and 4689. Of course, it's also critical to note that these AI tools aren't only a way for an employer to surveil its workers. Job replacement due to AI is a massive concern for our members, particularly those in call centers and customer service companies all over the world have replaced workers like me with AI without an opportunity to change roles within the organization. And we aren't just a line on a budget. We're people with families and mortgages and health problems. And we need employers to consider its workers more as they look at AI tools like what is implemented in Amazon Connect. If AI is going to be as devastating to certain jobs that proponents expect, or perhaps even hope for, then notice and retaining or retraining should be the bare minimum that employers offer affected workers. So please Support Senate File 4576. Thanks again to the committee for hearing my testimony and to support Senator Maya Quaid for these bills, and I'm happy to take questions.

Good afternoon, Madam Chair and members of the Senate Labor Committee. My name is Lauren Schadhorst with the Minnesota Chamber of Commerce. We appreciate the challenges around designing a regulatory framework for a seemingly ubiquitous yet novel technology we all lovingly refer to as AI. As we've begun reviewing the trio of bills before you seeking to regulate the use of automated decision systems and electronic monitoring tools. In particular, we want to flag with some top line thoughts a few areas where overly broad or inconsistent definitions may inaccurately or unintentionally capture practices or tools in ways that could create existential challenges for business and negatively impact employees, customers and citizens. We welcome the opportunity to thoughtfully work with you to ensure bills are structured correctly to achieve stated goals, while limiting the negative consequences due to a full lack of understanding of what is and is not exactly AI, an automated decision system or electronic monitoring tool. To that first question, what is AI? You can see in the three bills before you today it's defined almost identically, but in Senator Make Way's Build in Commerce and AI is defined as the meaning given in United states Code Title 15, Section 9401, while in other bills like Senate File 4575, 4573 and 4509, it is not defined the same. These definitional inconsistencies are an example of the need for a more thoughtful approach to regulating an entirely new technology and its potentially far reaching impacts on our lives and how business is conducted in our state. To be clear, the Chamber's position is that AI should be addressed at the federal level to avoid a contradictory patchwork of different state rules. Just last week the White House announced a framework for Congress to consider developing AI policies which will hopefully move the needle for that federal framework to materialize before states get too far out of line. Before Minnesota proceeds down the path of state level regulation, agreement on a definition of the term across the enterprise of state government seems like a good first step. For another example in the Senate Commerce Committee, the focus on AI has been pricing of consumer goods. In those committee, Senator McQuade seeks to regulate the use of an automated decision system. Like the bills before you today. While you can read how the bills before you define an ADS here in the bill before Senate Commerce, an ADS does not have the same technical definition. We doubt that a small business operating in our state would be able to identify how the AI tools they use to today would be prohibited, allowed or even regulated under any of these proposals. Other companies with more sophisticated or complex applications of AI may also be challenged to know what is or is not allowed under these proposals. The definition is broad enough to encompass routine behavioral and transactional information used in business analytics that shouldn't be considered AI or might even predate it. Across retail, hospital, hospitality, logistics, manufacturing and technology sectors, employers rely on automated tools to manage inventory, prevent fraud, administer loyalty programs, respond to demand, and structure compensation systems. These are foundational components of modern commerce. It will be challenging for employers to determine what types of data inputs or operational factors qualify under these new standards. And what if all of these bills happened to to pass this session? How would a reasonable person be able to comply with multiple different definitions of what AI is or is not? When key terms vary from bill to bill, compliance becomes a moving target. With heightened disclosure and procedural obligations, regulatory uncertainty increases legal exposure, slows innovation and discourages investment. To the bills before us as drafted, some of the requirements are unprecedented could limit the effectiveness of tools already used in situations for legitimate purposes or unintentionally be included in the scope will be difficult to operationalize across a workforce or multiple locations. And all, of course, with penalties, penalties and more penalties that will certainly create exposure for technical or inadvertent violations. The Minnesota Chamber continues Surrey continues surveying our members to get a better sense of how are using AI and related tools and how these bills would interfere, augment, support or challenge their operations. And it's not clear to us if it is 8 out of 10 employers at this point in time. We believe a more thorough approach is needed before policymakers establish a regulatory framework that is specific to Minnesota, particularly given the need for a federal framework and a comprehensive understanding of AI's uses and benefits for our economy. For these reasons, we oppose moving these bills at this time and look forward to working with this committee as this process develops. Thank you, Madam Chair.

Sure. Thank you, Madam Chair. Good afternoon everyone. My name is Dalton Danielson, Fiscal Policy Director with the Minnesota Business Partnership. I stand today in opposition to the bills before you, but let me start with a point of agreement. Technological change is real and workforce transitions deserve thoughtful, meaningful policy discussions. Employers across Minnesota are already investing heavily in training, reskilling and responsible deployment of new technologies. However, the bills before you today take a concerning and potentially very damaging approach. In practice, these bills create rigid, punitive requirements that do not reflect how businesses actually operate in competitive, fast moving markets. Technology adoption is often incremental, experimental and responsive to global competition. These bills will discourage innovation and slow productivity growth, potentially harming not only Minnesota's employers, but Minnesota's workforce as well, limiting their opportunities for future new opportunities, new job fields, whatever it may be. Additionally, the penalties in this bill are also significant. For example, incentive file 4576 employers could face back pay, civil fines and loss of eligibility eligibility for state grants, loans and tax incentives for five years. For some employers, this could mean thousands, tens of thousands, or even millions of dollars. That is not a targeted remedy. That is an overcorrection that risks pushing investment, expansion and job creation out of Minnesota altogether. Furthermore, the penalties become more concerning given the lack of standard definitions. As the Chamber mentioned, without standard definitions, employers will be operating in constant ambiguity, concerned that any move they make could jeopardize their entire business model given the low threshold for what's defined as a covered employer. We are talking about a lot of businesses here in Minnesota more broadly. And to conclude, we urge the Legislature to proceed with great caution before unnecessarily restricting artificial intelligence. AI is a general purpose technology that is already improving productivity, safety and job quality across industries. Premature or overly prescriptive regulation risks freezing innovation before we fully understand where the technology will ultimately land and this could place Minnesota employers and employees at a competitive disadvantage relative to other states. For these reasons, we respectfully urge the Committee to oppose these bills and and continue conversations with stakeholders.

Thank you and good afternoon, Madam Chair and Members, My name is Melissa Heising and I am Legislative Director of the Minnesota AFL CIO. I'm here to testify in support of Senate files 4680, 89, 4686 and 4576 as an important step towards common sense regulation of how AI is used in the workplace and its impact on workers. Working people are the experts on how to do how to do their jobs well and safely and they must have a voice in how AI is deployed in the workplace. On Senate file 4689 we appreciate the strong foundation this proposal establishes to regulate automated decision systems in the workplace. It includes critical protections, advance notice to workers, limits on algorithmic wages, strong appeal rights and clear prohibitions on using these systems to violate labor, labor rights, civil rights or safety laws. Those are important steps, but to ensure these systems are truly accountable, several improvements could be made. First, the bills could require impact assessments, not just disclosure, to evaluate accuracy bias, disparate impacts and risks to worker rights before systems are deployed, including independent third party review, regular updates and a stop use requirement if discriminatory outcomes are found. The bill could also include and should also include anti retaliation protections for workers who raise concerns or refuse to follow the output of an automated decision making system when the worker believes in good faith that the output would cause harm, discrimination or otherwise violate law or professional licensing requirements. On Senate file 4686, the surveillance bill. Workers need clear protections to ensure monitoring does not undermine their privacy, their dignity and their rights on the job. This bill contains important safeguards including advance notice before monitoring tools are deployed, ensures workers can access their data and appeal employment decisions influenced by monitoring, and places meaningful limits on surveillance such as banning monitoring in private private spaces, prohibiting off duty surveillance and restricting technologies like facial or emotion recognition except for narrow security purposes. Those are important protections however, similar to the first bill, we think it could go further to ensure monitoring tools are used responsibly by requiring impact assessments to evaluate privacy risks and potential harms again before the systems are implemented. On Senate File 4576. This bill recognizes that AI and automation are already transforming workplaces and workers deserve fairness, transparency and time to adapt when technology displaces their jobs. We believe this bill could be strengthened to ensure workers are truly protected by extending the timeline, doubling the notice and transitional employment period from 90 days to 180 days so workers have real time to prepare. The thresholds could also be broadened so more workers are covered, including part time workers, and so smaller layoffs triggered by automation would still require notice. Second, we should ensure the transition actually leads to quality jobs. Training programs should be worker centered and connected to proven pathways such as joint labor management training partnerships and registered apprenticeships. And finally, enforcement provisions could be strengthened, including stronger penalties, private right of action and anti retaliation protections. Thank you.

Thank you, Madam Chair. The Amendment is the 8th. This goes through and kind of updates what we were looking for. So it updates the reduction in equivalent hours from 25% to 10%. And for the whole department or department equivalent, we instruct the commissioner to develop templates for covered employees to provide notice of technological displacement. It also updates the transitional employment period that they're exempt if they have if they're transitioned to a position with same or similar benefits or better benefits.

If we're eliminating an entire department, but we're going to move everybody in that department to just a different place in the company, and we're going to train you to be part of that. Then they're exempt from that. We are adding in on page three after line six that the commissioner must consider the size of the employer and the severity of the violation when assessing penalties. And then there is a report that is added to the chairs and ranking minority members of labor, and then it adds an effective date.

Yes, Madam Chair, you would like to do. Yes.

Madam Chair, I think this may be probably the most simple bill, I think, of the three, and I do just want to name that. I think that a lot of the job loss we might see or job displacement from AI will also be people who will never go into the job market. I know that for recent graduates or people who are transitioning, it's very, very tough to get into entry level positions. And so I think, you know, there's a body of work to maybe be done around that. But I think this is a really good step because we are seeing job displacement due to AI, and while we've seen job displacement in the economy before due to advancements in technology, this is much broader in scope. And so I'm really glad that we're taking a look at this now and able to move forward.

Thank you, Madam Chair. I guess first I'd like to know the committees this has to go to yet and kind of the path of where we're looking for.

Thank you, madam Chair.

Well, there is Attorney General Enforcement. I believe in all three of these. I'm not 100% sure if that automatically sends it to Judiciary. Invoking Attorney General Enforcement doesn't always do that. And I think 831 is the jurisdiction of this committee.

So, anyway, I don't know if this one needs to go to Judiciary automatically, but the other two, I believe would.

Thank you, Madam Chair, to Senator McQuaid. Thank you for bringing this. It's a very important issue, so I appreciate that. My intentions for both of us would be to protect workers. I'm just concerned with the amount of regulations and the speed of this bill, with all the stuff that's in it, and we're just running out of time. I guess my first question would be as far as the work that you, you put into this so far, the stakeholders, have you worked with the businesses with DLI and just kind of how did you put this together and then maybe use some other states to model after.

Thank you, madam Chair. And Senator Doornick, I am very happy to give a lot of the credit to my author in the other body who spent the whole interim working with states and stakeholders and, and national partners. I think what you're looking at In Senate file 4576 is a combination of two states, New Jersey and New York, who have some of these provisions. So I have been learning along the way with my counterpart in the other body, but this has been a lot of his work is what you're seeing before you. Senator Dornick, thank you.

Senator McQueen.

Yes.

Senator Joyner, Madam Chair, thank you. Senator McQuait. So it's 50 or let's see, 50 or more. Is that your intent? And then is how many businesses would that affect in Minnesota? Would you know that? And then I've heard this from a couple testifiers that are representing business that are very concerned about this. So other than I know, I think It's Representative Greenman. Is that who's been. No. With the House. I'm just wondering how much you had mentioned there was some business interaction, but again, both testifiers were very concerned about this.

Thank you, madam Chair. And Senator Doornick. For a covered employer, it's 50 or more full time employees. I don't have the number off the top of my head of how many many that affects, but it's many. And it's Representative Gottfried in the House who's been working on it. And I think it's important to know for this, for this bill, the job displacement bill, this really is about both giving notice to people who are going to be displaced, having a transitional employment period, and then there's incentive to transition those folks to other places within that business if possible. And so, you know, I'm not really sure how much that would affect businesses in Minnesota beyond saying, hey, if you're going to lay a bunch of people off in a certain part of your business, we would like to know that and we'd like to know people being displaced.

Thank you, Madam Chair. Thank you, Senator McQuade, for the bill. I would agree that if folks are

we can do is make sure they're given notice in that transitional period. But I was just curious where the timelines came from. I think a minimum of 90 days is good in terms of notice. And then I see that additionally for the transitional employment period, do we see similar notice periods in any like, like other area of law is that notice

Just curious if it's where that kind

see, we see similar language anywhere else.

Thank you, manager. Thank you, Senator Mover Bane for the question. I believe this is modeled after and I would look to. No, I wouldn't. I believe it is modeled after a time parameter that does exist somewhere else in law. If it's not somewhere else in law, then it is modeled after two other states who have this notice period in law or both.

Thank you, Madam Chair. And no specific question, but I'm just gonna make a statement or comment about and this can be about all three. I won't say it three times, but I think the concern I have is that, you know, statement made by Ms. Hysing, you know, that employees need to be a part of the process when these things are going to happen. But when I listen to the, the testimony. I don't believe business has been involved in being a part of the legislation that's going to be telling them what they must or must not do. So I think it is incredibly important that if we're going to move forward on legislation that is going to dictate how businesses operate, and if we're going to abide by the premise that employees need to be a part of the process of things that are going to impact them, then we must also follow that same premise if something is going to impact employers. And from the testimony I heard today, I just do not believe they have been engaged enough in this process. So I hope the intention is that we are not expecting to move this into law this year, that we are going to be. I mean, I really appreciate the conversation because I do share a number of concerns about AI in general, but this isn't something that we should be acting on, just kind of a knee jerk reaction. It's something that's going to take some time to engage with a number of people, get all the perspectives and be willing to take those perspectives into account in the legislation that we're going to end up with as our end product. So I hope, I hope our process is, we're beginning a conversation that we know is going to take some time, more than two months that we have left in this legislative session.

Thank you, Senator Rarick. Senator Mayquid, would you like to respond? Sure.

Thank you, Senator. Members, any further discussion on Senate File 4576 as amended. And I'll just take the opportunity before we move this bill, Senator, that I actually, I also appreciate your Comments Senator Rarick and I appreciate the collaborative nature with which you've approached this work in general. My fears actually are probably on the other end. I share some of the concerns that were voiced by the Minnesota AFL cio. I'd like to see some of these, what I feel are just very basic standards in how people should be treated just strengthened up a little bit. So I'm a little, I have this fear that 10 years down the road, we're going to be looking back at this time of us being in this legislature and just realize how we had an opportunity in this time period right now, this session, maybe next session. But things are moving so quickly, so much faster than we are used to moving at the legislature that I just, I'm more worried that we are we are already behind, but I'm worried that we're going to fall further and further behind this technology and that people are going to suffer as a result. So I think it's imperative that we get moving. Not seeing any further comments. Would someone like to move that Senate File 4576, as amended, be recommended to pass and re referred to the Committee on State and Local Government. Okay, thank you. On that motion. All in favor, please say aye. Aye. Any opposed? The motion passes. All right, next we're moving to Senate File 4686, and I believe there's an amendment on this as well. Senator McQuade.

Madam Chair, we also have this listed as an A2.

Y.

a quick will you tell me what the difference between the A1 and the A2?

Madam Madam Chair, members.

to the Workforce Development Fund. Thank you, Madam chair, for the A2amendment for Senate file 4686. One thing that I do want to point out is on line 1.17 we've taken out the word assist on the definition of automated decision system so that it is used currently says to assist or replace human discretionary decision making. That is used and we're changing it to use to partially or fully replace human discretionary decision making. That is a really important point. I think assist might be a little too vague or too low. So fully or partially replacing human decision making system. Then you'll see some just updated tightening of language. So instead of on page eight, line 12, workers who are off duty, we said outside of regularly scheduled work hours, we tighten up. A lot of this is tightening up language or making sure there's really clear definitions or really clear words. And then we are directing the penalty to go to the workforce development fund instead of the individual. That's the amendment. Happy to talk anymore about the specific of the bill.

We're on 4686.

Thank you. On electronic monitoring. Yes. You know, I was, I actually appreciated Nico the Testifier giving the experience of working at the Minnesota Department of Revenue. That's our own enterprise in that I had no idea that that level of monitoring and surveillance was going on of our state employees and with a company to whom I don't know for sure that that data isn't going to, to the entire company of Amazon and being used in other ways that we wouldn't that aren't related to completing the job and being evaluated for the job. So a really, really good example of before that gets deployed to the rest of the enterprise, let alone any other part of our workforce, to make sure that if employees are being electronically monitored, it's for these reasons. It is to ensure quality of goods and services, facilitate compliance with laws and regulations, protect the health, safety and security of workers or the security of the employers, facilities or computer networks. I think those are great reasons to have electronic monitoring or at least acceptable reasons to have electronic monitoring. Outside of that, we're veering really far into over surveillance of our workers and then using digital tools to do that with no guardrails currently in law around what that data can be used for, sold for or traded for. And in this increasingly digital world, we shouldn't have to give up our privacy in order to have a job. And so this bill, I think takes really good steps to rein in unacceptable uses of electronic monitoring. And I also, there was a bill that Senator Lucero actually gave me a few days ago about not having an employer implant electronic monitoring inside of a person. And I said, well, boy, will you love my bill on electronic monitoring because that is also included in here as well. So that's line 8.24. Physically implant devices that collect or transmit worker data or installed in personal items of clothing or accessories. So this is, I mean, it sounds it sounds weird when you say it out loud, but there's two members of different caucuses who've introduced bills around this. Right. And so this is stuff that is happening or proposed to happen, and we really just want to make sure that workers aren't giving up all of their privacy of their entire life just by going to work because of the electronic monitoring tool that company has chosen to use.

Thank you, Madam Chair. I was just wondering with Dolly if one of their representative would come up and I'd like to ask a question or two and then just see how they're doing with the bill.

Thank you, Madam Chair. Thank you. So I guess being the enforcement, I'm just, I know you guys are really busy with the many different things that we've assigned you to investigation, and we appreciate you and the work you do. So thank you for that. Just with this and the task that you're going to have, I'm just wondering one, you know, maybe is this going to require more full time employees, some more finances to help enforce or just kind of address those first?

Thank you. Chair Josiah Moore, Legislative Director at Department of labor and Industry. Thanks, Senator Doornick. Generally, we are working on a fiscal note for this bill and the other two that are in front of the committee today. So I don't have specifics on the costs, but I can just say generally that whenever DLI is assigned enforcement, new enforcement, we typically project costs and project staff costs associated with that enforcement.

So you're saying at this time you don't have it all? Is that what you said?

Thanks, Chair. Yeah, we don't have specifics, but some general information about what we would generally project.

Please go ahead.

Senator dornick.

Okay. Senator Rare.

Senator McQuade, thank you, madam Chair. And thank you, Senator Rarick, for the question. Yes, we have in on 7.9. So we have what they can, what employers can use electronic monitoring for. So ensuring quality of goods and services, facilitating compliance with laws and regulations, protecting the health, safety or security. And then if you look over at the electronic monitoring tool definition on 2.4, it just means any system, application or instrument that facilitates the collection of data by means other than direct supervision by a person. So direct supervision is also always available to every employer.

And documentation of misconduct or failing to meet benchmarks or standards, all of that is always available to employers as well.

Thank you, Madam Chair, and I appreciate that. But yet, as we have moved into this remote and hybrid world, many people who are working in their homes, you do not have that direct supervision. So you're relying on some of these technologies to monitor, to make sure that they're continuing their workforce. Like maybe that, yes, they're actually at their computer working. So I hope we're not taking away some of that ability, because not everyone has that direct supervision happening.

Thank you, Madam Chair. And Senator Wearik, you know, I think you're talking about something different than what electronic monitoring would accomplish. And so if Somebody's not at their computer and they're not doing their job. Right. They're maybe not responding to emails or not showing up for meetings or not available when they're supposed to be available. Those are always benchmarks that are observable by a person, even if it's not like in that person's home. Like you can see when your colleague isn't responding to the emails or finishing their part of the process project or logging into the system to update the whatever.

All of those benchmarks are available and tell us how people are doing at their job. And none of that requires you to be monitoring keystrokes. Right. Because when you're given a job, you have deliverables that you have to give as part of your work and if you're not meeting those, that's not going to be trackable by how many keystrokes did you enter? What was your facial expression while you were on the phone with somebody?

And so, so the things that you're talking about always available to employers and would never want to take away the ability for employers to measure outcomes for employees. But I want to take away is the ability for employers to over surveil their employees and insert irrelevant things as meaning of how well are you doing at your job? Like how does your face look when you're on the phone with somebody or you know, did you go to the bathroom for five minutes? Right. Like those are, those are not necessary to talk about somebody accomplishing their job deliverables.

Senator RICK yeah, thank you, Madam Chair. I mean I appreciate that I'm not in my reading through this. I understand I haven't had as much time as I would have liked to fully grasp this, but I think, you know, we've heard before that oh, don't worry, this isn't covered. I mean we the prime example, we can go back again. This was federal, but Obamacare, oh, you can keep your doctor. Oh, don't worry. This is going to lower rates and then come to find out that wasn't true. So I think we need to, you know, because it's the intent that it's in there, that's what we want. The words on the page matter and dealing with all the folks that are impacted by this and making sure they agree that the language is doing what you're saying is doing is important. And if we have the groups that represent businesses that come forward and say we don't agree, I hope we're listening and trying to make changes so that it addresses what we say it's addressing and not actually having other impacts that were unintended.

No.

Yes, Madam Chair. So this on 4689, on that line 1.17, you're seeing that same replacement of assist to partially or fully replace human discretionary decision making. One place, I guess I really want to note, because it's a big and I think important Change on line 1.7 of the amendment. So page nine, lines 11 and 12 in the bill. We're essentially saying that an automated decision system cannot terminate a worker. That actually has to be the decision of a human being. And electronic monitoring can collect the things to present to a human. You know, I'm going to recommend this person for termination, or here's all the stuff that might lead you to that conclusion. But the ultimate decision to terminate a worker does have to be a human being. This also puts the funds from the violations in the workforce development fund, also adds in the language about the commissioner considering the size of the employer and the severity of the violation when assessing penalties, and then puts in an effective date of January 1, 2027.

So this bill is about automated decision systems. And to be very candid about what automated decision Systems are. It's when a system is taking in information and then making decisions without a human on behalf of that system. I should say based on the inputs that it's getting. To be very clear, automated decision systems are email filters, firewalls, antivirus software, identity and access management tools, calculators, data databases, data sets, or any other compilation of data. When it comes to automated decision systems, the bill requires pre use notice. So the employer has to let employees know that this is going to happen. What sets of decisions are the automated decision system allowed to make and what inputs are going to be put in to make those decisions and allow employees to read that in a very easy, digestible way and not like, you know, our terms of service. When you log into social media and then it lays out what needs to be contained in the pre use notice, it sets out requirements for record keeping, record requests. And it's really important because if employees, if there is a decision that is made and an employee is affected by that, like let's say they're, you know, placed on a pillow because of an automated decision system, but it's for something that that person never did or said, or maybe they weren't even there at work that day because AI is notorious for hallucinating. The right to access their own information and then request human review of the decision. There's post use notice requirements as well. Standalone communication, easy to understand. I think I talked about right to access, right to appeal. And then of course there's the enforcement and anti retaliation language as well. That is the automated decision system. I do want members to just think about the difference between an automated decision system and maybe the way we would use AI or algorithms. So imagine that when we introduce bills, instead of going to a person and a person saying this is what committee it goes to an algorithm, read our bills and just decided what committee it went to. We would want the right to say, actually that was a mistake. Why did it think that I understand and fix that mistake. That's what this bill essentially does, is it makes sure that there can be human reviewers for any decisions made and that the decisions made are limited to workplace issues and not of great consequence, like firing you without a human determining that, or taking in information that is not necessary to the job, like your disability status, your health status, your religious or political beliefs in order to make any other determinations about you or your employment.

Thank you, Madam Chair. Thank you. Senator McQuaid, again, just want to. We agree there's some things that need to be done. So thank you for coming here. I was wondering if I could bring up Lauren Shot Horse to ask a few questions. Would that be okay? Yes, if she's okay to come up.

Thank you. Thank you for your testimony. And it's a lot for us to try to digest and there's a lot with the three bills and stuff. I was wondering if you could help us with some, maybe some examples how it affect and some of the things that maybe you're concerned with in the bill.

Sure. Mr. Madam Chair, Senator Dornick, Lauren Shadhorst with the Minnesota Chamber of Commerce. You know, in looking through a variety of the bills as defined, the scope could include a number of technologies like geolocation tools, video surveillance, time tracking software, photo optical systems that are far more than high risk AI monitoring. A restaurant using a basic shift scheduling app or a trucking company using GPS fleet tracking for payroll purposes would both be covered.

In full. I'm looking through some of the examples. A small trucking company using fleet trucking technology to log delivery routes have the same notice, consent and data access infrastructure as a large employer would be covered using AI driven performance monitoring. Those are not the same things, but they would be treated the same under the bill.

Yes, Senator McQuade, it's just a technical correction. The affirmative consent piece was deleted in the A2amendment. So I just want us to be clear as we move forward.

When we're looking at scheduling software, task routing tools, customer queue systems, basic performance dashboards, boards could be covered. Exclusions do not protect the broad middle ground of common enterprise software that uses statistical modeling or optimization logic to surface recommendations to managers. So a workforce scheduling platform that uses demand forecasting to suggest shift assignments is not a database or a calculator. It's a computational process issuing simplified output used to assist in an employment related decision could fall square within the definitions under the bills. So the point being is these are broad, sweeping definitions and without a clear understanding as to what is being utilized already that might not be fully contemplated in the bill could be covered. And you know, in looking at the future of this piece of legislation or these multiple pieces of legislation, trying to see what level of specificity might be necessary or to confirm whether or not these things should be covered, at all. And the drafting today, I don't think hits that balance.

Sure. Two parts. If you look at lines 1.20 and 1.21, it does talk about what's not an automated decision system. Compilation of data is one of them. So if you're looking at, you know, about how many people did we have, Fourth of July weekend, come through the store to guess how many shifts we need? That's a compilation of data. That's not an automated decision system. And to be really clear, this doesn't prohibit automated decision systems. So a small trucking company and electronic monitoring that uses tracking for its fleets, they just have to notify their drivers that they are tracking that, tell them what they're tracking and what they're then using it for. That's, I don't think, a huge burden. And I think workers have a right to know when they're being tracked, whether it's driving a truck or working in a store or sitting at a computer and working for the Department of Revenue.

Appreciate that. So I guess, last question. Does this affect Uber Lyft Doordash, and how does that interact with them at all?

Thank you, Madam Chair. Are you Senator Duron? Are you talking about automated decision making or for.

Yeah.

Thank you, Madam Chair. So automated decision systems would be impacted, but they would just be required to notify them what kind of system they're using, what they're using it for, what data they're putting into it, and that's fine. Does that make sense?

Senator Doornick, that's it. Unless Ms. Schouthorse wanted to respond to Ms. Sch.

I would just say we would disagree about the burden that this would place. It's incredibly burdensome, confusing definitions, applications beyond what might be intended in the scope of this bill.

I was actually just going to say the same thing. You know, Senator Mae Quaid, when you said you don't think it's burdensome, you know, have you checked with nfib? Have you checked with any of these other groups to see if they believe it's burdensome or not? I mean, I think that's part of the problem we have here. Too many times we as legislators think we know the answer to something and we haven't checked with those who were putting this on to see if they agree that it's burdensome. And I also think again, with many of these last two, especially, you know, when we're talking about the penalties that are involved, I think we should be working with those groups to see just what kind of a impact. I mean, that might be a, we think that that fine could be proper for a very large business, but that could be crippling to a small business that makes a mistake in their notification or something. So I think those are things we all have to really think about as this moves forward. But I do have one just technical question I think, for our staff to help me understand these last two bills. The definitions are, if not virtually the same, very, very similar, but we're putting them into two different spots of statute. And is that, is there a reason we're doing that rather than just creating the same spot for this in reference to the bill instead of having it, if they all pass, having it in multiple places.

with the author, just, you know, if

So I think that was just the

But yes, we would keep track of that and not want them to conflict, et cetera.

Follow up, Senator.

Yes, follow up.

Thank you for that follow up or other comments, questions. Yes.

Senator MAQUIN thank you, Madam Chair. Senator RICK It's a really good question. I'm glad you asked it. Part of the answer to that question is there are state legislators from across the country who are working across party lines. So I work with Republicans and independents and other states and Democrats and other states on legislation that, you know, attempts to get our hands around the way that consumer facing tech and AI has been rolled out in society with no regulation. And so the hope is that we can pass enough state laws where there's a critical mass, where, you know, there's a lot of Fortune 500 companies headquartered in the state of Minnesota.

And so it's not just companies who are in Minnesota, but we got headquarters here too. And if that's going to be the state of play, then that's going to be the state of play for all, you know, Best Buys or all three Ms. Or all targets. The second part, I will say is tech is, is finally having their reckoning. And in the last two two days alone, social media companies have been found liable for addictive and harmful behavior and owe money to those victims. And so I think we're finally in a place and I think the bipartisan work in state legislatures across the country is evidence of the fact that we can no longer let big tech go unregulated in this way. And in particular this surveillance of our life has been sold to us as unique, this thing, because it's going to be so helpful, it's going to be so great for you. And also we get all this data about you and we get to do all of these things to you. And we really have to break that apart. And again, like, I don't think Minnesota companies are the ones who are saying, oh, I would love to, like, know the, you know, psychological state of my employees and what they might do, you know, in a week. But buying technology that allows them to shift schedule efficiently also gives them that information. And that just shouldn't be.

Thank you, Madam Chair. One last follow up then. And I don't know what your other bills are, but if this is like this one specifically is meant to kind of go after the technology companies, why isn't it being written to them and their products rather than to the businesses and putting the penalty on the businesses? Why aren't we going after the tech companies with this legislation?

Thank you, madam Chair and Senator Barrick. I, you know, as, as we were having this conversation, I was like, I'd be really happy to write a bill regulating, you know, not just ADS and employment, but like, how ADS systems are designed and sold to companies in the state of Minnesota. And I think there, there is some work happening on, like, what are the ingredients that are included in some of these electronic models, monitoring systems and automated decision systems as well. So that, you know, I think a lot of companies probably don't even get the full scope of the information when they're said, here's this like, efficiency product. And so it's a really good idea and I think, I think it's worth doing. The issue before us though is that as we heard from our testifier at the Department of Revenue, it's already happening. And so we do need regulations in place for these systems as they've already been deployed. But I do think you're right that there maybe is a bigger bill to be done about these systems and how they're designed and how they're unleashed in the public.

Yeah. Thank you. So no more questions, but just one follow up to that. And it's the, as you just stated, you think a lot of the companies, they're getting these technologies from a company and they probably don't even know, you know, two thirds of what's in it. And then my worry is we're gonna, that company's gonna get this piece of software, whatever it is, and then they're going to be on the hook for the fines rather than the tech companies, even though they didn't even know it was there, they weren't using it, somebody's gonna find out it's there. It's, it's in the cloud or wherever the heck it is, and it's the company that's gonna be punished rather than the tech company that, you know. So that's. I hope we can keep considering those things as this moves forward.

Madam Chair. Yes, thank you, Madam Chair. And I really appreciate that. One of the things we were just talking about, I don't know what day it is sometime this week, was potentially requiring disclosure to companies at the point of sale for our contract for an electronic monitoring system or automated decision system of everything that it includes so that companies have the information before them of the full scope of what is possible so they can make those decisions with that full information that companies are required to disclose that. And so you are on the right track and I'll probably bring you that amendment before it comes on to the bill.

Thanks.

So moved.