US DOL Issues Guidance on AI and the FLSA, FMLA, PUMP Act, and More!


On April 29, 2024 the U.S. Department of Labor issued Field Assistance Bulletin (FAB) No. 2024-1, Artificial Intelligence and Automated Systems in the Workplace under the Fair Labor Standards Act and Other Federal Labor Standards, to provide guidance on the Fair Labor Standards Act, the Family and Medical Leave Act, and other workplace laws in light of the increasing use of artificial intelligence and other automated technologies in the workplace.

Issued pursuant to President Biden’s AI Executive Order (discussed in our October 31, 2023 blog post), which directed the Secretary of Labor to provide guidance on the FLSA and AI, the FAB emphasizes that federal laws continue to protect employees regardless of the tools and systems used in the workplace and regardless of where the work is performed. AI and other automated technologies can perform many functions, including tracking work hours, measuring performance, setting work schedules, assigning tasks, and performing certain human resources functions. The FAB identifies specific laws and the issues that may arise under each, as follows:

Fair Labor Standards Act. The FLSA requires employers to pay employees for all hours worked, including overtime. The FAB notes that there are areas of concern with regard to both hours worked and wage payment.

  • Hours worked. AI and certain monitoring tools can measure and analyze when workers are “active” or “idle,” such as through keystrokes, mouse clicks, website browsing, or camera presence. The FAB warns there must be human oversight of these systems, since these tracking systems are not determinative of hours worked, and any errors by AI in categorizing work time will result in violation of the obligation to pay for all hours worked.
  • Break time. Employers may use AI to predict or automatically populate time entries for breaks based on schedules, prior time entries, business rules and other data. But these entries may not necessarily be accurate.
  • Waiting time. When an employee is “engaged to wait” to work, the time is considered work hours for which they must be paid, as compared to when they are “waiting to be engaged,” which is not work hours. The FAB asserts that AI may have the ability to assign tasks and set work schedules automatically, which may change throughout the day based on changing data. Potential issues can arise when employees are waiting for their next task or an updated schedule – if there is not sufficient time for the employee to use for their own purposes or if they must remain nearby, they are “engaged to wait” and must be paid. Employers must accurately account for time spent waiting for the next task.
  • Work performed at multiple geographic locations. Some employers use location-based monitoring to track employees and determine if they are “working.” This may cause problems if the system fails to account for work performed in different locations – such as when an employee must pick up supplies before reporting to a job site or perform tasks offsite, as well as travel time between worksites.
  • Calculating wage rates. Employers may use AI to calculate or adjust wage rates based on a variety of data and metrics, which can include fluctuating supply and demand, customer traffic, geographic location, worker efficiency or performance, or type of tasks. Employers must ensure that these systems are accurately calculating the wage rate, particularly where the employee is paid multiple wage rates.

The Family and Medical Leave Act. The FMLA provides employees with unpaid, job-protected leave for certain family and medical reasons. The FAB identifies particular issues that may arise with the use of AI:

  • Processing leave requests. AI may be used to determine eligibility, calculate leave entitlements, or evaluate whether leave is for a qualifying reason. However, there may be errors in calculating hours worked, which could result in incorrect eligibility determinations, or in calculating the amount of leave available.
  • FMLA certifications. Employers are strictly limited in what information may be requested to support a request for FMLA leave. AI may request more medical information than allowed. It may also improperly deny leave when a certification deadline is missed, without allowing for circumstances that permit additional time.
  • FMLA interference and retaliation. Systems that track leave use may not be used to target employees for retaliation or discourage the use of leave. FMLA leave may not be automatically counted for purposes of disciplinary or other employment actions.

Providing Urgent Maternal Protections for Nursing Mothers Act. The PUMP Act provides nursing employees with reasonable break time and space to express breast milk for a year following the child’s birth. The FAB identifies the following concern:

  • Scheduling breaks. Automated technologies that limit the frequency, timing, and length of breaks required for pumping may violate the law, since these may vary depending on each employee’s needs. Additionally, systems that penalize employees for failing to meet productivity standards will violate the law if they do not take pump breaks into account. It is also a violation if the system requires employees to make up pump breaks or reduces future work schedules because of these pump breaks.

The Employee Polygraph Protection Act. The EPPA prohibits private employers from using lie detector tests on applicants and employees, except in certain industries under limited circumstances. The FAB notes the following:

  • Determining when an employee is lying. Some AI technologies can take certain physical measurements – such as eye movement, voice analysis, or micro-expressions – to determine if an individual is lying. This type of measurement is covered by the EPPA.

Retaliation. An employee’s complaints to the DOL about violations of the FLSA or certain other laws are protected from retaliation. The FAB raises certain retaliation issues:

  • AI systems cannot be used to penalize or discipline employees for engaging in protected activity under these laws.
  • AI systems cannot be used to conduct workplace surveillance for the purpose of identifying protected activity under these laws or targeting employees who have made complaints, and then taking adverse action against employees for engaging in such protected activity.
  • Although not within the purview of the DOL, the FAB further notes that AI systems cannot be used to predict the likelihood that certain locations will unionize.

This guidance from the DOL was followed on May 16, 2024 by its Artificial Intelligence and Worker Well-being: Principles and Best Practices for Developers and Employers, as discussed in our May 16, 2024 blog post.