- A July 2026 lawsuit against Meta alleges its AI performance-scoring system penalised employees on protected leave during layoffs, extending discrimination liability beyond hiring into workforce reduction.
- In Mobley v. Workday, a federal court ruled AI vendors can be held directly liable as employer “agents” under the ADEA, piercing the contractual firewall that vendors had assumed protected them.
- New York City’s Local Law 144, the EU AI Act’s August 2026 enforcement deadlines and active state proposals in New Jersey mean employers now face compounding AI compliance obligations across multiple jurisdictions at once.
The July 2026 lawsuit against Meta Platforms targets something most AI employment litigation has not: layoffs, not hiring. The complaint alleges Meta’s internal performance-scoring systems penalised employees on protected leave because they could not accumulate the activity data the algorithms measured. That fact pattern, if it holds, would expose employers to discrimination claims at every point where an AI system touches a workforce decision.
Meta Lawsuit Targets Layoff Algorithms
The 26 current and former Meta employees who filed suit in July 2026 allege the company used keystroke monitoring, activity-tracking data and AI token-usage dashboards to score, rank and select employees for termination. Employees on protected leave had no way to accumulate the activity inputs the system measured. The complaint claims Meta did not neutralise those inputs for leave-takers, exclude them from the scoring cohort, or pause for the individualised review required under federal and state law, leaving the system to generate disparate outcomes based on sex, pregnancy and disability.
Most AI employment litigation to date has focused on hiring tools. This complaint targets ongoing workforce management and reduction, a much broader surface area of employer liability. Federal anti-discrimination law, including Title VII of the Civil Rights Act and the Americans with Disabilities Act, applies to all employment decisions, not only hiring. The Meta lawsuit tests whether that principle holds when the decision-maker is an algorithm.
Vendor Liability Comes Into Focus
Mobley v. Workday has been running in parallel and is equally consequential. The plaintiff alleges that Workday‘s AI hiring tools discriminated on the basis of age, disability and race by factoring in data points including employment gaps and medical-related leave that correlate with protected characteristics. A federal court allowed the federal claims to proceed earlier this year, rejecting Workday’s dismissal arguments. Critically, the court held that Workday could be treated as an “agent” of the employers using its tools, which opens the vendor itself to direct liability under the Age Discrimination in Employment Act. That ruling punctures the assumption that AI providers can shelter behind a contractual firewall while employers absorb the legal risk.
A separate class action, Kistler et al. v. Eightfold AI Inc. filed in California in January 2026, adds a data privacy dimension. The complaint alleges the AI hiring platform scraped personal data on more than a billion workers, secretly scored applicants and discarded low-ranked candidates without human review, framing those practices as violations of the Fair Credit Reporting Act and California’s Investigative Consumer Reporting Agencies Act. The Equal Employment Opportunity Commission’s position throughout has been consistent: AI used in hiring is a selection procedure, and employers remain liable for discriminatory outcomes even when a third-party system produces them.
The EU AI Act Deadline
The EU AI Act entered into force on August 1, 2024. Its high-risk AI obligations, the category employment-decision systems fall under, were originally set for August 2, 2026, but the EU’s Digital Omnibus deferred that deadline to December 2, 2027. For a fuller picture of how obligations are structured, see our analysis of its four-tier risk framework.
The Act’s extraterritorial reach is the detail most relevant to non-European employers: any company whose AI outputs affect users within the EU must comply, regardless of where the company is headquartered. AI systems used in employment decisions fall squarely within the high-risk category, which carries penalties of up to €15 million or 3% of global annual revenue, whichever is higher, separate from and lower than, the €35 million/7% tier reserved for prohibited practices like social scoring or manipulative AI.
US State Laws Add Local Liability
New York City’s Local Law 144 requires annual independent bias audits of automated employment decision tools, public disclosure of audit results and advance notice to candidates. It is already in effect and has shaped how employers in the city document and disclose their use of AI screening tools, a compliance pattern that Connecticut’s SB 5 looks set to extend further.
New Jersey has active proposals adding to the picture: S3263, introduced in February 2026, covers video interview consent; S4279, introduced in May 2026, covers algorithmic impact assessments. Neither has passed, but their progression reflects how quickly state-level compliance requirements are accumulating.
Audit, Document, Oversee
Across these legal fronts, three practical obligations keep appearing. Vendor due diligence means more than asking whether a tool is “bias-free”, employers need to see bias testing procedures, data privacy controls and documentation of how the system reaches its outputs. Human oversight matters because no hiring or termination decision should be fully automated; the Meta and Workday cases both turn partly on whether meaningful human review occurred. And documentation, covering AI system configurations, training data, bias testing results and decision logs, is increasingly the difference between a defensible position and an indefensible one.
Regular adverse impact analyses, using the four-fifths rule as a baseline, can surface disproportionate effects before litigation does. Transparency obligations are now explicit in several jurisdictions: candidates must be told when AI is being used, and some laws require offering an alternative evaluation path. Employers operating in the EU or in countries with works council requirements, including those subject to Belgium’s Collective Bargaining Agreement No. 39, face additional obligations to consult employee representatives before deploying AI workforce tools.
The financial exposure from non-compliance is compounding across multiple jurisdictions simultaneously. That combination makes inaction a more costly option than it was 12 months ago.
