Artificial intelligence has moved quickly from a back-office experiment to a daily workplace management tool. Employers can now use software to record computer activity, measure idle time, analyze keystrokes, review customer communications, score applicants, rank workers, flag “low productivity,” and recommend discipline or termination. For some employers, these tools promise efficiency, consistency, safety, and better use of data. For workers, however, they can feel like constant surveillance, especially when the system is opaque, collects more information than necessary, or reduces complex human performance to a score. Michigan House Bill 5579 of 2026, introduced as the proposed Responsible Artificial Intelligence Security for Employees Act, responds directly to that tension by placing proposed limits on electronic monitoring tools and automated decision tools used in employment. ¹
The most important point for Michigan employers and employees is that the bill is proposed legislation, not current law. As of the current bill history, HB 5579 was introduced in the Michigan House in February 2026 and referred to the House Committee on Economic Competitiveness. ¹ The bill would create a new statutory framework governing when employers, third parties, and service providers may use electronic monitoring or automated decision systems in the workplace. Its structure reflects a growing national debate: whether workplace AI should be treated merely as a management tool or as a regulated system that can affect wages, schedules, discipline, hiring, promotion, privacy, and job quality.
The bill’s reach is broad. It would apply not only to traditional employees, but also to applicants and independent contractors who perform work for remuneration for an employer operating in Michigan. ¹ That matters because modern workplace AI is often deployed before a person is hired and continues after the person begins work. A résumé-screening model may decide who receives an interview. A scheduling algorithm may decide who receives desirable shifts. A productivity dashboard may decide who is coached, disciplined, or terminated. By defining “covered individual” to include employees and applicants, and by defining employment-related decisions to include wages, benefits, hours, scheduling, performance evaluation, hiring, discipline, promotion, termination, job content, work assignment, productivity requirements, and workplace health and safety, the proposal attempts to regulate the full employment life cycle rather than only one stage of employment. ¹
The bill distinguishes between two major categories of technology. An “automated decisions tool” would include computational processes, including machine learning, statistical modeling, data analytics, or artificial intelligence, that produce simplified outputs such as scores, classifications, or recommendations used to substantially assist or replace human discretion in employment decisions.¹ An “electronic monitoring tool” would include systems that collect data about a covered individual’s activities or communications by means other than direct observation, including computer, telephone, camera, radio, wire, photoelectronic, or other electronic systems. ¹ In practical terms, this language would reach many common workplace systems, including applicant screening tools, productivity dashboards, time-tracking platforms, keystroke or mouse-movement monitoring, GPS tools, call-center analytics, warehouse performance metrics, and certain safety or quality-control systems.
The proposed restriction on automated employment decisions is especially significant. HB 5579 would generally prohibit an employer from using an automated decision tool to make an employment-related decision. ¹ The bill does include a limited exception allowing use of an automated decision tool to screen large volumes of job applications for candidates who meet set hiring criteria or to assess candidates based on job skills. ¹ Even that exception should be read carefully. The bill does not appear to give employers broad permission to let an algorithm make final hiring, firing, promotion, discipline, wage, scheduling, or evaluation decisions. Instead, the proposal draws a line between using technology to help process large applicant pools and using technology to make consequential decisions about a worker’s job.
That proposed line is important because automated decision tools can make employment decisions look more objective than they really are. A score may appear neutral, but it depends on the data used to train the system, the variables selected by the vendor, the employer’s assumptions about what counts as “productive,” and the way the output is interpreted by managers. If a tool rewards constant keyboard activity, it may undervalue planning, mentoring, problem-solving, customer care, or physically demanding work performed away from a computer. If a tool relies on historical employment data, it may reproduce past inequities. If a tool ranks applicants based on patterns found in prior “successful” employees, it may unintentionally screen out qualified applicants who do not resemble the existing workforce. HB 5579’s proposed restrictions reflect the concern that employment decisions should not be delegated to systems that workers cannot understand, challenge, or correct.
The bill would not ban all electronic monitoring. Instead, it would allow monitoring only for specified purposes. Under the proposed language, an employer could use an electronic monitoring tool to allow an employee to accomplish or facilitate an essential job function, monitor production processes or quality, periodically assess performance, ensure compliance with labor or employment law, protect health, safety, or security, administer wages and benefits under limited circumstances, or accomplish another business-operation purpose determined by the Michigan Department of Labor and Economic Opportunity. ¹ This framework does not treat monitoring as inherently unlawful. Rather, it requires a close connection between the monitoring and a legitimate work-related purpose.
The proposed law also emphasizes narrow tailoring. An employer using an electronic monitoring tool or automated decision tool would have to provide written notice, obtain written consent, ensure that collected data is accurate and up to date, allow workers to correct inaccurate data, use the tool in a narrowly tailored manner, use the least invasive means possible, apply the tool to the smallest number of covered individuals necessary, collect the least amount of data necessary, and use the tool no more frequently than necessary. ¹ These requirements would move workplace monitoring away from a “collect everything first and decide later” model. The bill’s approach is closer to data minimization: decide why the employer needs the tool, collect only what is needed for that purpose, and avoid surveillance that is broader, more frequent, or more intrusive than the business reason requires.
One of the most notable provisions would prohibit collection of employee data while the employee is off duty. ¹ That issue is becoming more important as remote work, employer-issued devices, mobile apps, wearable technology, and cloud-based systems blur the line between work and personal life. A worker may use the same phone for work and family communications. A remote employee may work from a bedroom, kitchen table, or shared household space. A driver, nurse, technician, or delivery worker may carry a device that tracks movement throughout the day. The bill’s off-duty limitation recognizes that employer technology can extend beyond the workplace unless there are legal boundaries around when monitoring begins and ends.
The bill also identifies categories of data that employers could not collect through permitted tools. These include health, medical, lifestyle, and wellness information; protected characteristics such as race, disability, sex, gender identity, sexual orientation, pregnancy-related information, national origin, ancestry, veteran status, and similar characteristics; certain workplace activities; communications; device usage; geolocation information; audio-video and sensor data; facial, emotion, gait, or voice recognition information; automated decision inputs and outputs linked to a covered individual; and online information such as private social media activity or internet protocol addresses. ¹ The breadth of this restriction is striking because it would affect many tools currently marketed as productivity, safety, engagement, or workforce analytics products.
The proposal’s treatment of productivity information is particularly important. Many AI workplace monitoring products are built around efficiency data, idle time, response times, application usage, call volume, movement, location, or task completion. HB 5579 would allow certain performance-related monitoring for limited purposes, but it would also restrict collection of productivity and efficiency information in ways that may require careful interpretation if the bill advances. ¹ Employers would need to examine whether a tool truly measures a necessary job function or instead collects broad behavioral data that is only loosely connected to performance. A worker’s value is rarely captured by a single metric, and the bill appears designed to prevent employers from converting every digital trace into a disciplinary record.
The bill would also prohibit monitoring in private areas such as bathrooms, locker rooms, changing areas, breakrooms, smoking areas, employee cafeterias, lounges, lactation spaces, prayer areas, and similar locations. ¹ It would also prohibit certain monitoring in an employee’s residence, personal vehicle, or property owned or leased by the employee. ¹ This provision is a response to the reality that remote and hybrid work can bring employer systems into intimate spaces. Even if an employer has a legitimate reason to protect data or verify work activity, the proposed law would recognize zones of privacy that should not become part of an employer’s surveillance architecture.
Another important provision would prohibit tools equipped with facial, gait, voice, or emotion recognition technology. ¹ Emotion recognition, in particular, has become controversial because it often claims to infer attention, honesty, stress, fatigue, engagement, or attitude from facial expressions, tone, posture, or other biometric signals. In employment, those inferences can be especially risky. A worker’s face, voice, movement, or affect may be influenced by disability, culture, language, fatigue, illness, neurodivergence, trauma, or ordinary personality differences. When a system converts those signals into judgments about performance or trustworthiness, it can create a false sense of scientific certainty. HB 5579 would take a categorical approach by barring these recognition technologies in covered tools rather than merely requiring employers to disclose them.
The proposed law would impose substantial notice obligations. Employers would have to provide written notice to covered individuals subject to electronic monitoring or automated decision tools and obtain written consent. ¹ The bill would also require workplace posters, advance notice before implementation, notice in job postings, notice on the employer’s website, direct notice to applicants, and accessible formats that account for language and disability needs. ¹ The notice would have to give covered individuals the ability to opt out. ¹ If a covered individual opted out, the employer could not use the tool to make employment-related decisions about that individual. ¹ This opt-out structure would be a major compliance issue because many employers deploy monitoring or analytics systems uniformly across departments or job categories. If workers can opt out, employers may need parallel processes for evaluation, scheduling, discipline, hiring, and productivity assessment.
HB 5579 would also require impact assessments before use. Under the proposed bill, an employer would need an independent and impartial third party to assess the tool’s objectives, algorithms, data, cybersecurity vulnerabilities, and potential biases, including discriminatory outcomes based on race, gender, or disability. ¹ The assessment would need to examine the attributes and modeling techniques used, whether those techniques are scientifically valid, whether they may function as proxies for protected characteristics under Michigan’s Elliott-Larsen Civil Rights Act, whether the tool may create disparate impact, whether it may limit accessibility for people with disabilities, and whether it may negatively affect privacy or job quality, including wages, hours, and working conditions. ¹ This is one of the bill’s most consequential features because it would require employers to understand and document how a system works before using it on workers.
The proposed timing of the assessment is also demanding. The bill would require an assessment one year before implementation or, for tools already in use when the act becomes effective, within six months after the effective date. ¹ Employers would also have to conduct or commission annual assessments for each year the tool remains in use. ¹ Within sixty days after completion, the employer would have to submit the assessment, or an accessible summary, to the Department of Labor and Economic Opportunity for inclusion in a public registry and distribute it to covered individuals who may be subject to the tool. ¹ This would make AI and monitoring governance more public than many employers are accustomed to. Vendors may also face pressure because the bill would require service providers to allow employers access to documentation necessary for the assessment. ¹
The documentation requirements would likely change vendor contracting. Many employers purchase monitoring or AI systems from third-party providers and may not have access to training data, model specifications, validation studies, version histories, or detailed explanations of how outputs are generated. HB 5579 would require employers to retain documentation about design, development, use, and data, including data sources, technical specifications, individuals involved in development, historical use data, and historical versions of the tool. ¹ If a vendor cannot or will not provide that information, the employer may be unable to comply. In practice, employers would need to negotiate contract terms requiring transparency, audit rights, data-security commitments, retention limits, breach cooperation, and support for worker notices and correction rights.
The bill also contains data retention and data transfer limits. Employers would have to retain collected data for no more than three years after the purpose for using the tool is achieved, unless a collective bargaining agreement provides otherwise. ¹ If the employer does not use specific data, it would have to delete that data immediately. ¹ The employer could not sell or license covered individual data, including deidentified or aggregated data. ¹ Sharing data with the state or a local unit of government would also be limited to specified circumstances, such as providing information to the department, complying with law, or responding to a court-issued subpoena, warrant, or order. ¹ These provisions target the secondary market for workplace data and the risk that information collected for employment purposes could later be reused for unrelated commercial, governmental, or disciplinary purposes.
The breach provisions are among the most aggressive parts of the bill. If a security breach affected data collected through an electronic monitoring tool or automated decision tool, the employer would have to secure the systems, mitigate harm, certify corrective steps, notify affected covered individuals within forty-eight hours after discovery, and notify the department and attorney general. ¹ The bill would also require extensive protections for affected individuals, including ten years of paid identity theft protection and insurance, comprehensive credit monitoring, dark web monitoring, breach alerts, credit freezes, fraud remediation, Social Security number monitoring and reissuance costs, and bank fraud and financial transaction monitoring. ¹ The proposal would also require a post-breach third-party audit to confirm vulnerabilities have been fixed. ¹ These obligations would create powerful incentives for employers to limit the amount and sensitivity of worker data they collect in the first place.
The proposal is also significant for unionized workplaces. HB 5579 states that it would establish minimum standards and would not diminish employees’ collective bargaining rights. ¹ If employees are covered by a collective bargaining agreement and the employer intends to use an electronic monitoring or automated decision tool; the employer would have to provide notice and an opportunity to bargain overuse of the tool to set or influence wages or other terms and conditions of employment. ¹ Before collective bargaining, the employer would have to provide necessary information, including data collected, impact assessments, and information about data breaches. ¹ This aligns with broader federal labor-law concerns about electronic surveillance and automated management practices, including the National Labor Relations Board General Counsel’s position that intrusive monitoring can interfere with employees’ rights to engage in protected concerted activity.⁵
The business community has already raised concerns. The Michigan Chamber of Commerce has described HB 5579 as a proposal that could impose significant administrative and legal obligations on employers, including limits on automated decision tools, notice and consent requirements, impact assessments, union information obligations, and new legal exposure. ³ Those concerns should not be dismissed lightly. Employers use monitoring and AI for many legitimate reasons, including workplace safety, cybersecurity, quality control, wage administration, regulatory compliance, fraud prevention, and operational efficiency. A poorly drafted or overly rigid law could make it difficult for employers to use beneficial tools or respond quickly to operational needs. At the same time, the existence of legitimate uses does not answer the core question: how much monitoring is too much, and who decides?
Supporters of the legislation frame the bill as a guardrail rather than an anti-technology measure. Reporting on the bill’s announcement emphasized concerns about tools that record screens, track mouse and keyboard movement, monitor breaks, and generate productivity reports that can affect discipline or other employment outcomes. ² The practical problem is not simply that employers collect data. It is that workers may not know what is collected, how long it is kept, who sees it, whether it is accurate, how it is interpreted, whether it can be challenged, and whether it is being used to make consequential decisions. HB 5579 would attempt to answer those questions through notice, consent, opt-out rights, correction rights, assessments, public registry obligations, and legal remedies.
For employers, the most practical first step is to inventory existing systems. Many organizations do not think of their ordinary software as “AI” or “monitoring.” A timekeeping system, help-desk dashboard, applicant tracking system, call-center analytics tool, route optimization platform, productivity plug-in, learning management system, fraud detection tool, or scheduling program may collect worker data or generate recommendations that influence employment decisions. Employers should identify what data each system collects, why it is collected, who has access, how long it is retained, whether the vendor uses it for product improvement, whether the system produces scores or rankings, and whether managers use those outputs in employment decisions.
Employers should also separate measurement from decision-making. A system that helps a worker perform an essential function may present different risk than a system that recommends discipline. A tool that tracks machine output for quality control may be less problematic than a tool that records every screen or analyzes private messages. A dashboard that gives employees feedback may be different from one that automatically lowers performance ratings. HB 5579’s focus on employment-related decisions means employers should pay close attention to how data moves from collection to consequence. The legal risk usually increases when monitored data becomes the basis for pay, scheduling, promotion, discipline, termination, or hiring.
Human oversight would be essential if legislation like HB 5579 becomes law. The U.S. Department of Labor’s AI best-practices roadmap similarly emphasizes meaningful human oversight for significant employment decisions, transparency with workers, worker input, protection of labor and employment rights, training, and data security.⁴ Human oversight should not mean rubber-stamping an algorithmic recommendation. It should mean that a trained manager understands the tool’s limitations, reviews relevant context, considers whether the data is accurate, gives the worker a meaningful chance to respond, and independently decides whether an employment action is justified. A human being must remain accountable for the decision.
Employees, for their part, should understand that proposed regulation would not necessarily eliminate all monitoring. Even under HB 5579, employers could still use certain tools for essential job functions, quality, compliance, safety, and limited performance assessment. ¹ The likely change would be that employers would need to justify, disclose, minimize, assess, and document those tools. Workers would gain more information about what is happening and more ability to challenge inaccurate data. That shift matters because workplace monitoring is often invisible. A worker cannot correct a false productivity score, challenge an unfair algorithmic ranking, or object to excessive surveillance if the worker does not know the system exists.
The bill would also create legal remedies. A covered individual aggrieved by a violation, or a person acting on that individual’s behalf, including a labor organization, could bring an action for damages, injunctive relief, or both. ¹ A prevailing plaintiff could recover economic damages, noneconomic damages, costs, and attorney fees. ¹ A violator could also face a civil fine of up to $500, and prosecutors or the attorney general could bring an action to collect the fine. ¹ Although the civil fine itself may appear modest, the private right of action, attorney-fee provision, public registry, impact assessment obligations, and breach remedies could create meaningful litigation and compliance exposure.
The deeper issue is workplace trust. Monitoring can improve safety and efficiency when it is targeted, transparent, and tied to real operational needs. It can also damage morale when employees feel reduced to data points or punished for metrics that do not reflect actual work. AI can help identify patterns, but it can also miss context. It may not know that a nurse spent extra time comforting a patient, that a legal assistant paused typing to think through a filing issue, that a technician solved a problem through experience rather than speed, or that a remote employee was productive while not constantly moving a mouse. A workplace that uses AI responsibly should be able to explain why it uses the tool, what the tool measures, what it does not measure, and how people can correct mistakes.
HB 5579 should be watched closely by Michigan employers, technology vendors, unions, employees, and applicants. If enacted in its current or similar form, it would require a substantial rethinking of workplace monitoring, productivity tracking, automated screening, vendor contracts, data security, employee notices, applicant communications, performance management, and labor relations. Even if the bill is amended or does not become law, it signals where the policy debate is heading. Employers that depend on opaque monitoring or algorithmic management should expect more questions from workers, unions, regulators, and courts. Employers that build transparent, narrowly tailored, human-centered systems now will be better positioned for whatever legal framework emerges.
The proposed Michigan approach reflects a broader shift in employment law: data about workers is no longer treated as a harmless byproduct of doing business. It can shape careers, wages, schedules, opportunities, and reputations. When AI systems collect that data and transform it into scores or recommendations, the stakes become even higher. HB 5579 would place meaningful limits on that process by requiring purpose, notice, consent, minimization, independent assessment, correction rights, privacy protections, security obligations, and accountability. Whether the bill becomes law or not, it presents a clear message for the modern workplace: productivity technology should serve legitimate business needs without turning every worker into a permanently monitored data profile.
This article is for general informational purposes only and is not legal advice.
Contact Tishkoff
Tishkoff PLC specializes in business law and litigation. For inquiries, contact us at www.tish.law/contact/. & check out Tishkoff PLC’s Website (www.Tish.Law/), eBooks (www.Tish.Law/e-books), Blogs (www.Tish.Law/blog) and References (www.Tish.Law/resources).
Footnoted Sources
1- Michigan House Bill No. 5579, 103rd Legislature, 2025–2026 Regular Session, introduced version, “Responsible Artificial Intelligence Security for Employees Act.” https://legiscan.com/MI/text/HB5579/id/3373691
2- Colin Jackson, Michigan Public Radio Network, “Democrat-led bill looks to regulate AI workplace monitoring in Michigan,” February 23, 2026. https://www.michiganpublic.org/politics-government/2026-02-23/democrat-led-bill-looks-to-regulate-ai-workplace-monitoring-in-michigan
3- Michigan Chamber of Commerce, “Michigan lawmakers propose strict limits on how employers can use AI to monitor employees,” February 25, 2026. https://www.michamber.com/news/michigan-lawmakers-propose-strict-limits-on-how-employers-can-use-ai-to-monitor-employees/
4- U.S. Department of Labor, “Department of Labor releases AI Best Practices roadmap for developers, employers, building on AI principles for worker well-being,” October 16, 2024. https://www.dol.gov/newsroom/releases/osec/osec20241016
5- National Labor Relations Board, “NLRB General Counsel Issues Memo on Unlawful Electronic Surveillance and Automated Management Practices,” October 31, 2022 https://www.nlrb.gov/news-outreach/news-story/nlrb-general-counsel-issues-memo-on-unlawful-electronic-surveillance-and