Employers be warned: If you rely on artificial intelligence (“AI”) or automated systems in the workplace, you must ensure proper oversight. While recent technological advancements have provided employers with ways to increase productivity and improve work efficiency, the U.S. Department of Labor’s Wage and Hour Division (“WHD”) makes clear in a recent Field Assistance Bulletin (“FAB”) that the use of such technology is not a substitute for responsible human oversight.
This timely FAB, entitled Artificial Intelligence and Automated Systems in the Workplace Under the Fair Labor Standards Act and Other Federal Labor Standards, highlights the potential shortcomings of AI and other workplace technologies, and identifies risks and compliance challenges associated with the use of such technology under the Fair Labor Standards Act (“FLSA”) and other federal labor standards. Although the FAB’s practical guidance is minimal, the overarching sentiment is that employers must oversee the use of AI and other technologies, and the manner in which it is used, to ensure it does not result in violations of federal labor laws and standards.
Specifically, the FAB addresses compliance challenges that may arise under various federal employment statutes, including:
- FLSA: Determining “hours worked,” break time, waiting time, offsite work hours, and wage calculations;
- Family and Medical Leave Act (“FMLA”): Processing and certifying leave requests;
- Providing Urgent Maternal Protections for Nursing Mothers Act (“PUMP Act”): Complying with break time requirements and anti-retaliation provisions;
- Employee Polygraph Protection Act of 1988 (“EPPA”): Determining whether the EPPA applies (yes, this might apply to you!); and
- Anti-retaliation provisions: Ensuring use of technology does not result in prohibited retaliation under WHD-enforced federal laws.
Read on for more details concerning the compliance risks that arise with AI and other workplace technologies in these areas. As a reminder, the laws referenced within the FAB do not replace or supersede state or local laws providing greater protections to employees, including the FLSA and PUMP Act which may not apply to certain California employers.
Compliance with the FLSA
Tracking Hours Worked
The FLSA requires employers to pay employees for all “hours worked,” regardless of the level of productivity or performance, so long as the employee is “suffer[ed] or permit[ted] to work.” While employers once largely relied on timecards and spreadsheets for timekeeping purposes, automated timekeeping and monitoring systems are rapidly replacing these manual processes to track not only hours worked, but also an employee’s activity and productivity. The problem arises when employers rely on this data to determine the “hours worked” by an employee for purposes of federal wage and hour laws.
For example, the FAB highlights certain AI programs and monitoring systems that measure and analyze real-time worker activity and productivity by tracking keystrokes, mouse clicks, website browsing, eye movements, or webcam presence to determine whether an employee is actively working. Compliance issues arise when these tools incorrectly categorize an employee’s work time as non-compensable based on its analysis of an employee’s activity and productivity, which can result in a failure to pay wages for all hours worked. The FAB warns employers that these productivity-measuring tools are not determinative of “hours worked” under the FLSA, nor are the metrics produced a substitute for the analysis of whether an employee performed “hours worked” under the FLSA.
Monitoring Break Time
Likewise, employers face compliance risks when using automated or predictive timekeeping systems to account for break time under the FLSA. The FLSA counts short breaks of 20 minutes or less as compensable hours worked, whereas longer breaks are non-compensable when an employee is completely relieved of duty, and the break time is long enough to enable the employee to use the time effectively for their own purposes.
Automated timekeeping systems that make predictions or auto-populate time entries based on an analysis of prior time entries, regular shift and break times, and other data may pose a risk when “smart” entries automatically deduct meal breaks and other longer break periods from an employee’s compensable work hours, even when the employee is not completely relieved of duty. Without appropriate human oversight, a system that automatically deducts breaks could result in timekeeping and wage payment violations under the FLSA. Accordingly, the FAB reminds employers that the use of AI does not relieve an employer’s responsibility to ensure they accurately record break time and properly compensate employees for all hours worked.
Accounting for Waiting Time
Employers who utilize automated scheduling and task assignment programs may run afoul of the FLSA’s requirement to compensate employees for hours worked when an employee is “engaged to wait,” i.e., required to be available and prepared to perform work. In contrast, time for which an employee is “waiting to be engaged” is not considered compensable hours worked.
Potential compliance challenges arise when an automated system fails to account for hours worked when an employee is not actively performing a task but is “engaged to wait” for their next assignment. For example, the FAB notes that hotels often use automated systems that assign tasks to housekeeping workers based on their availability and other factors, including their real-time activities. Additionally, warehouses and distribution centers often utilize algorithmic systems that analyze workflow and provide workers with real-time instructions to perform tasks.
Since both these systems utilize real-time data of a worker’s activity to generate output, this often results in unpredictable reassignments of tasks or adjustments in schedules, during which workers must wait for an update on their next assignment or schedule of assignments. The FAB warns that, during this time, an employee is “engaged to wait” if the employee is not provided with sufficient time to use for their own purposes, is not completely relieved of duties, or is expected to remain near their workstation without a set time to report back to work. Accordingly, employers must ensure that wage records and payments accurately account for time during which an employee is “engaged to wait,” in addition to the time taken to complete tasks, to comply with the FLSA.
Determining Hours Worked at Multiple Geographic Locations
The FAB reminds employers that the location of work is not determinative of whether an employee has performed “hours worked,” so long as the individual is “employed” during the time they are “suffer[ed] or permit[ted] to work.” Employers using geolocation software, or location-based monitoring, to track employees should be aware of potential risks that arise when an employee performs work-related tasks off the job site, which results in a longer workday than their scheduled shift or presence at a particular worksite.
Compliance issues occur when a system records what it determines to be an employee’s “hours worked” based solely on their location or automates the clock-in/clock-out process when an employee enters and leaves the job site. This automated recording can result in a worker being compensated only for the time spent at the worksite, even if the worker is performing work away from the worksite (e.g., an employee purchases supplies at a store before arriving at the worksite, or unloads supplies offsite after leaving their worksite). The FAB warns that employers risk minimum wage and overtime violations under the FLSA when systems fail to account for time spent working offsite or time spent traveling between worksites. Accordingly, employers must ensure all hours worked are accurately recorded and compensated, regardless of the location in which the work is performed.
Calculating Wages
The FAB also reminds employers to ensure that workers are paid in accordance with applicable federal and state wage laws when using AI or other technologies to calculate and/or determine workers’ wages. Employers should pay particular attention to systems that independently calculate and determine a worker’s pay rates based on variable data, such as supply and demand, geographic location, efficiency, or performance; systems that automatically recalculate and adjust pay rates throughout the day; and automated task assignment systems that determine a worker’s individual tasks based on various factors. The FAB notes that when a worker is paid different wage rates based on different metrics, the employer must ensure the differing rates are properly calculated into the regular rate of pay. For workers paid multiple rates, the FAB suggests that the employer and employee may agree to compute overtime pay at 1.5 times the hourly rate in effect at the time the work is performed.
Compliance with the FMLA
Employers who use automated systems to handle FMLA leave requests also may encounter compliance challenges, specifically, when the systems are used to determine eligibility, calculate available leave, evaluate whether leave is qualified, or request supporting documentation for a leave request. For example, the FAB notes that timekeeping and automated systems may miscalculate the hours worked to determine eligibility for leave, incorrectly dictate the days to be counted against an employee’s leave entitlement, or “test” for eligibility more frequently than permitted under the FMLA, resulting in leave being impermissibly denied.
Additional risks arise when the use of an automated system results in overbroad requests for disclosure of medical information, or automated penalties for “missed submission deadlines” when the employee was permitted extra time for submission. Finally, the FAB reminds employers that retaliation and interference are prohibited under the FMLA and cautions employers to ensure that automated systems do not analyze leave as a negative factor in employment actions or fail to provide benefits to employees on FMLA leave where such benefits are provided to other employees on similar types of leave.
Although these types of compliance errors can occur with human decision making, the FAB warns that automated system errors may result in violations that apply consistently across the entire workforce. Accordingly, employers must exercise responsible human oversight when deploying automated systems to handle and process leaves.
Compliance with the PUMP Act
Employers also should be mindful of their obligations under the PUMP Act, especially when utilizing automated systems to track and determine employee work hours, set schedules, assign tasks, manage break time, or assess worker productivity. The PUMP Act requires employers to provide nursing employees with reasonable break time and space to express breast milk at work for up to one year after the child’s birth. Although an employer and employee can agree to a certain break schedule based on the employee’s need to pump, an employer cannot require the employee to adhere to a fixed schedule that does not meet the employee’s need to pump. Additionally, any agreed-upon schedule may be adjusted based on a change in the employee’s pumping needs.
The FAB warns that automated scheduling or timekeeping systems that limit the length, frequency, or timing of a nursing employee’s pump breaks would violate the reasonable break time requirement under the FLSA, as amended by the PUMP Act. Violations also may occur when productivity scoring or monitoring systems penalize workers who fail to meet quotas or standards due to pump breaks taken. Likewise, an automated scheduling system that requires workers to make up time taken for pump breaks or reduces a worker’s future scheduled hours because they took pump breaks would constitute unlawful retaliation under the FLSA. Accordingly, employers must ensure workers are provided with lawfully required breaks and that automated systems do not impose adverse actions on employees for exercising their rights under the law.
Compliance with the EPPA
For employers who think the EPPA is inapplicable to their workplace technology, you might already be using technology covered under this law. By way of background, the EPPA prohibits private employers from administering lie detector tests to employees or to screen applicants, except in limited industries and under prescribed conditions. Under the EPPA, a lie detector is defined to include “a polygraph, deceptograph, voice stress analyzer, psychological stress evaluator, or any other similar device (whether mechanical or electrical) that is used, or the results of which are used, for the purpose of rendering a diagnostic opinion regarding the honesty or dishonesty of an individual.”
Given this broad definition, employers may already be using technology considered a lie detector test under the EPPA, if such technology is capable of analyzing an individual’s eye movements, voice stress, micro-expressions, or other bodily movements to suggest or detect deception. Employers who utilize this technology, and do not fall within the EPPA’s narrow exception, risk running afoul of the law. Accordingly, employers should ensure they fully understand the functions and capabilities of any AI technology used, particularly those with the ability to detect facial expressions and analyze body language.
Prohibited Retaliation
Finally, the FAB reminds employers that using AI or other technology to retaliate against workers constitutes a violation of federal labor standards. Automated systems implicated by the FAB include worker surveillance systems that target or monitor workers who are suspected of filing complaints or cooperating with WHD investigations, systems that predict the likelihood of workers unionizing in a particular location, systems that predict the likelihood of a worker filing a complaint or otherwise engaging in protected activity, or systems that use information regarding an employee’s protected activity or potential to engage in protected activity to make determinations about work schedules, leave, or assignments.
The FAB’s warning is not just theoretical: A group of Amazon warehouse workers in Missouri recently filed an unfair labor practice charge with the National Labor Relations Board, in which the company was accused of using “intrusive algorithms” in a workplace surveillance program to monitor and deter union organizing activities. Prior reports indicate that Amazon’s surveillance program included the monitoring of employee message boards, use of software to track union threats, and recruitment of intelligence analysts to monitor for “labor organizing threats.” This example emphasizes the importance of not only understanding the functions and capabilities of technology used, but also ensuring that technology is used in a lawful and responsible manner.
Takeaways
While technology has many potential benefits, employers must remember that they remain on the hook to ensure that their use of AI and other technology complies with all applicable law, which includes the FLSA and FMLA. There is no defense when the use of these technology without reasonable human oversight results in a violation of the law. To mitigate this risk, employers should, at a minimum:
- Ensure that appropriate personnel are fully informed on the functions, capabilities, and limitations of AI and other technology used;
- Train appropriate personnel on the compliance risks and “red flags” associated with the technology used;
- Ensure that the manner in which the technology is used comports with all applicable laws; and
- Maintain responsible human oversight to identify and mitigate system errors that can result in a violation of the law.
If you have questions relating to the DOL’s WHD Field Assistance or need assistance to address potential compliance issues regarding the use of workplace AI or other technologies, please reach out to the NFC Attorney with whom you typically work or call us at 619.292.0515.
