'Even if you establish that a particular system would produce discriminatory outcomes left and right, you have no way of sharing that information with other employees’
How can people laid off dueto decisions by AI prove in court that they actually were laid off with the technology playing a big part in the decision? That's the challenge that some Meta workers are dealing with right now, and one that HR professionals overseeing AI-assisted performance tracking or layoff decisions should watch closely.
In a recent case, a group of former Meta employees allege the company relied on AI tools, including a workplace assistant called "Metamate" and a productivity score built from keystrokes, screen activity, emails and browser history, to help select workers for layoffs. The plaintiffs say the scoring disadvantaged staff who had taken medical or family leave or who have disabilities.
Meta has told the court that humans made every decision involved in the roughly 8,000 layoffs announced earlier this year and denies that AI usage factored into who was let go, according to Reuters.
U.S. District Judge William Orrick last week declined to block the terminations of the 26 plaintiffs, saying he was bound to accept Meta's account because the workers could not produce evidence to counter it.
AI-driven decisions
Orrick wrote that employees challenging AI-driven decisions "were not in the rooms where it happened," Reuters reported. A hearing on a longer-lasting preliminary injunction, which could reinstate the workers, is scheduled for Aug. 24.
Like most U.S. workers, the plaintiffs are bound by arbitration agreements, meaning they cannot pursue a class action, put their case before a jury, or seek a large court settlement, according to Reuters. The lawsuit exists only because of a narrow exception in most arbitration pacts permitting workers to seek a court order blocking irreversible harm.
Christine Webber, co-chair of the civil rights and employment practice at Cohen Milstein Sellers & Toll, a firm not involved in the case, told Reuters that arbitration's confidentiality prevents findings in one case from helping other workers. "Even if you establish that a particular system would produce discriminatory outcomes left and right, you have no way of sharing that information with other employees," she said, according to Reuters.
A similar dynamic separates the Meta case from litigation against Workday, which faces claims its HR software filtered out job applicants based on race, age and disability; Workday denies the allegations. That case avoided the arbitration hurdle because Workday has no arbitration agreements with the applicants suing it, Reuters reported.
Sanchit Vir Gogia, chief analyst at Greyhound Research, said companies should stop assuming AI tools automatically make workforce decisions more defensible, noting there is "scant independent proof" that AI produces more accurate or lawful layoff choices, according to a report from CIO. His firm's analysis argues the burden sits with employers to show their tools work as claimed, not with workers to disprove it.
Meanwhile, Bayron Flores Tapia, an attorney at Phillips & Associates, said AI may automate hiring decisions, but it does not automate legal responsibility, Computerworld report. Courts have already signalled that outsourcing a decision to software does not shield an employer from liability, a point underscored in the ongoing Mobley v. Workday case, where a court allowed disparate-impact claims to proceed after finding a screening tool's near-total rejection rate for one applicant plausibly pointed to algorithmic bias.
What comes next
The plaintiffs' lawyers acknowledged in a joint statement that gathering evidence remains difficult and have asked current and former Meta employees with knowledge of the layoff process to come forward, according to Reuters. "Meta holds virtually all the relevant information," they said, Reuters reported.
Orrick's ruling at the Aug, 24 hearing will hinge on whether the plaintiffs can produce that evidence in the coming weeks.
A Meta spokesman said Tuesday the company had no further comment, according to Reuters.
U.S. technology companies announced 38,242 job cuts in May 2026, the highest monthly figure for the sector since August 2024, according to data published by outplacement and executive coaching firm Challenger, Gray & Christmas (CGC). For the third consecutive month, AI was the leading reason cited by U.S. employers for job cuts — accounting for 40% of all announced positions eliminated in May, up sharply from just 7% in January 2026.
