Six in ten managers now say they use AI to help decide who gets laid off, and nearly a quarter let the software make that call with no human review at all, according to a new survey of 1,000 U.S. managers. The finding marks a shift from AI as a hiring filter to AI as a workforce-reduction engine, and it is happening faster than most HR functions have built the governance to match.
The workforce cut just got a new decision-maker
The survey, conducted by ResumeTemplates.com among managers with direct reports at companies of 501 or more employees, found that 59% use AI to help decide who is laid off and 58% use it to help decide who is fired. Nearly a quarter, 24%, say they lean on AI for layoff calls “often or all the time.”
That is a meaningfully different use case than the AI screening tools HR has spent the last three years debating. Hiring algorithms filter candidates who are not yet employees. This is software weighing in on whether an existing worker keeps their job, with all the severance, unemployment, and reputational stakes that entails.
What the software is told to weigh
Managers who use AI in layoff decisions say the tools most often factor in performance and productivity scores (80%), attendance (57%), and salary or cost (42%). A smaller but still notable share direct the AI to weigh tenure (32%), sick days or medical leave (31%), paid time off (23%), and age (14%). The last two are the categories most likely to draw scrutiny from employment lawyers, since both intersect with protected classes and disability or age discrimination claims.
Oversight is inconsistent, and training is thin
The survey found real variation in how closely managers supervise the tools. Fifty-seven percent say they never let AI make a layoff decision without human review. But 43% say they occasionally take a hands-off approach, and 17% do so often or all the time. Ninety-one percent say they would override an AI recommendation they personally disagreed with, which suggests most managers still see themselves as the final check, not the AI.
The gap is in preparation, not intent. Thirty-eight percent of managers say they have never been trained on the ethical use of AI in HR decisions. Fifty-eight percent cannot confirm whether the AI tool their company uses has been tested for bias, and 23% confirmed outright that no such testing occurred. That leaves a majority of managers making or informing termination decisions without knowing whether the system they are relying on has been checked for the kind of disparate impact that regulators and plaintiffs’ attorneys look for first.
The survey also found a sharper edge to the automation: 13% of managers say they have already replaced a worker’s job with AI outright, and 44% have evaluated whether a role could be automated away entirely. That puts the layoff-decision question and the job-elimination question on the same desk, often decided by the same manager, using the same tool, in the same review cycle.
This follows the hiring side into the courtroom
Layoff decisions are arriving at the same moment AI’s role on the hiring side is already facing legal tests. HRTech has reported on how AI hiring and layoff tools are becoming employers’ biggest litigation exposure, as plaintiffs and regulators increasingly name the software vendor, not just the employer, as a defendant. A layoff tool that weighs sick days or age carries the same exposure in reverse: instead of screening someone out of a job before they are hired, it is weighing them out of one they already hold. The legal theory, disparate impact from an automated system, is the same on both ends of the employment relationship.
What it means for the HR leader
This survey lands alongside a broader reckoning already under way over AI’s role in workforce decisions. HRTech has covered the correction phase now hitting companies that leaned on AI-driven layoffs without building the audit trail to defend them later. The ResumeTemplates.com data suggests that correction phase is arriving into an environment where the underlying decision process itself, not just the headcount math, is under-governed.
For HR leaders, the immediate exposure is documentation. If a manager used an AI tool to weigh sick days, medical leave, or age in a termination decision, and that decision is later challenged, the employer needs to show what the tool weighed, why, and whether it was tested for bias before deployment. Right now, most cannot answer that with confidence. The survey also points to a training gap that predates the AI itself: managers were making judgment calls about performance and cost before AI arrived, and many were never trained on the legal boundaries of those calls either. AI has not created a new compliance category so much as it has made an old one harder to ignore.
What to do now
Three steps close most of the gap fast. First, require every AI tool used in reduction-in-force or termination decisions to carry documentation of its bias testing, including who ran it and when, before a manager can use it for that purpose. Second, set a firm rule on which factors the tool is permitted to weigh, explicitly excluding protected-class proxies like medical leave frequency and age, and audit the tool’s actual inputs against that rule. Third, train managers specifically on AI-assisted termination decisions, not just general AI literacy. Given that nearly two in five have had no training at all, this is the fastest lever HR has to close the exposure before the next reduction in force, not after.
Source: ResumeTemplates.com
