Seyfarth Synopsis: On April 29, 2024, the U.S. Department of Labor’s Wage and Hour Division released a Field Assistance Bulletin addressing the application of the Fair Labor Standards Act to use of artificial intelligence and other automated systems in the workplace.

Artificial Intelligence (AI) is seemingly ubiquitous.  By 2025, half of Human Resource departments are predicted to use AI in some capacity.  Employers are increasingly using AI-powered timekeeping systems or applications that can generate timecards, determine schedules and staffing, monitor performance, and process payroll (among many other benefits).  Against this backdrop, and pursuant to Executive Order 14110 (Oct. 30, 2023), the U.S. Department of Labor’s Wage and Hour Division (WHD) released a Field Assistance Bulletin (FAB) highlighting some of the compliance risks under the Fair Labor Standards Act (FLSA) associated with the use of AI in the workplace. 

Tracking Work Time

The FLSA generally requires that covered employees be paid at least the federal minimum wage for every hour they work and at least one and one-half times their regular rate of pay for each hour worked in excess of 40 in a single workweek.  One issue that often arises in wage-hour litigation is whether the employer had knowledge of any uncompensated work performed by an employee.  As noted here, courts generally hold that compensable work under the FLSA is work that employers require, know about, or should have known about.  

Traditionally, employers tracked time worked through either time clock machines or timesheets.  Increasingly, however, AI is being used for automated timekeeping in which software tracks when workers sign in and out of work and then determines an employee’s “active” and “idle” time.  For example, some AI-driven software monitors remote workers by taking screenshots of their computers at set intervals and collecting data, including keyboard activity and application use, to generate a timecard every 10 minutes throughout the day.  The software only records as time “worked” when the system detected “active” work, such as moving a mouse or a keystroke. Any periods of perceived inactivity are considered non-compensable idle time, which would not be reflected in the employees’ pay. 

WHD notes that this type of software may lead to wage-hour liability.  “An AI program that incorrectly categorizes time as non-compensable work hours based on its analysis of worker activity, productivity, or performance could result in a failure to pay wages for all hours worked.”  Although WHD does not provide examples of incorrectly categorized idle time, such examples may include AI-powered monitoring software failing to fully account for the time employees spend working away from their workstation; offline time spent thinking, strategizing, or resolving problems; time employees spend reviewing and researching hard copy documents or taking handwritten notes; or any offscreen engagement with clients or customers (among other “offline” activities).        

Monitoring Break Time          

The FAB also provides that employers must ensure that any AI-powered monitoring software recognizes compensable breaks.  Generally, short breaks of 20 minutes or less taken during the workday are generally counted as compensable hours worked, while uninterrupted breaks of 30 minutes or more are not.  See 29 C.F.R. §§ 785.18, 785.19. 

Traditionally, employers monitored break time through employees recording the start and end of their break.  By as WHD notes, “[s]ome timekeeping systems now incorporate AI to make predictions and to auto-populate time entries based on a combination of prior time entries, regularly scheduled shift times and break times, business rules, and other data.”   Moreover, automatic meal and break deductions are often a preferred default setting for employer timekeeping software, where a set period of time (e.g., 30 minutes) is automatically deducted from non-exempt employees’ pay for any given lunch period. 

According to WHD, such software may result in an FLSA violation, and provides the following example: “an employee usually takes a 30-minute unpaid meal break but skips the break on a particular day due to their workload.  Without appropriate human oversight, a system that automatically deducts the break from the employee’s work hours based on the employee’s past time entries could result in the employer failing  to properly record and pay the employee’s hours worked.”

Waiting Time

Not all idle time is compensable.  Instead, under the FLSA, idle time is compensable when employees are “engaged to wait,” but not when they are “waiting to be engaged.”     

Many companies have relied (and are relying) on AI tools and AI-driven software to manage staffing levels in a manner more finely tuned to changing demand, including having employees “on standby” and thus ready to work as needed.  For example, WHD found that “automated systems used by some hotels independently prioritize and assign tasks to housekeeping workers.  When a guest checks out of their room or requests service, these systems automatically delegate the task of cleaning that hotel room to a worker based on their availability and other factors.” 

The efficiencies achieved through use of this technology may also, however, pose wage-hour compliance risks.  “When the employee is not provided with sufficient time that they can use for their own purposes, is not completely relieved from their duties, or is expected to remain nearby their workstation and is not given a set time when to report back to work, they are generally considered to be ‘engaged to wait.’”  Therefore, WHD states that “employers must ensure that they accurately account for increments of time when the employee was waiting for their next assigned task,” unless the employee was sufficiently able to use his or her time effectively for personal activities.

Work Performed at Multiple Geographic Locations

Another area WHD identified where AI can result in unforeseen wage-hour risk involves the continuous workday rule, which provides that the period between the start and finish on the same workday of an employee’s principal activity or activities is generally considered compensable.  See 29 C.F.R. § 790.6.  Thus, travel from job site to job site during the workday is also generally compensable.  See id. § 785.38.

WHD found that some employers use location-based monitoring to track employees, which an automated system processes to determine whether an employee is “working.”  According to WHD, using such technology may inadvertently lead to wage-hour liability, and WHD provided the following example: “A system that records only the time the worker spent at the worksite as compensable work hours when the worker is performing work away from the worksite may fail to account for travel time between worksites or hours worked at other locations and may result in minimum wage or overtime pay violations.”

Calculating Overtime Compensation Owed

The amount of overtime pay due to an employee is based on the employee’s “regular rate of pay,” which is found by dividing the total pay for employment in any workweek (minus certain statutory exclusions) by the total hours worked during a given workweek.   

Issues sometimes arise when non-exempt employees are paid multiple wage rates (e.g., different hourly rates for different types of work).  In such a scenario, the employee’s regular rate of pay for that week is typically the “weighted average” of such rates.  See 29 C.F.R. § 778.115. 

WHD found that some AI-driven “systems have the ability to automatically recalculate and adjust a worker’s pay rate throughout the day . . ., which may result in significant different regular rates from one workweek to the next.  Similarly some automated task assignment systems have the ability to determine the number of types of tasks assigned to individual workers, based on a variety of factors and metrics,” and then pay workers at different rates based on tasks performed. 

WHD instructs that employers who use such technologies “must ensure that the different rates are properly calculated into the regular rate of pay[.]”     

Conclusion

AI and the use of automated technologies for scheduling, timekeeping, and calculating overtime may pose wage-hour risk.  Any such technologies, therefore, should be vetted to ensure that they do not lead to wage-hour liability.