Meta workers are suing over AI-scored layoffs. Construction firms running AI safety and productivity scores carry the same exposure.
Twenty-six former Meta employees say an internal AI scoring system flagged medical leave and disability accommodation as underperformance during May layoffs. Construction firms feeding wearable and productivity data into staffing decisions face the identical legal exposure — and some already have bias-audit obligations Meta didn't.
Twenty-six former Meta employees are suing the company, alleging that an internal constellation of AI systems — not human managers — built the list for its May 2026 layoffs, and that the scoring penalized workers on medical leave, disability accommodation, or pregnancy-related absence by reading their reduced activity as underperformance. Construction firms already running AI-scored wearables, productivity dashboards, and safety scoring carry the identical legal exposure — and in New York City and California, they already have bias-audit obligations Meta is accused of skipping entirely.
What does the lawsuit actually say?
The suit, filed July 14 in federal court in Oakland, comes from former employees across six states and Washington, D.C., tied to Meta's May layoffs that cut roughly 8,000 jobs — about 10% of its workforce. The complaint says Meta fed a mix of signals into the termination decision: its Metamate AI assistant, employee-trained "second-brain" agents, keystroke logs, AI token-usage dashboards, and performance calibration rankings. Plaintiffs argue the resulting productivity scores didn't account for approved medical leave or disability accommodations, so time out for a health condition read as a performance gap. They also claim Meta never bias-tested the systems, which they say violates disability, medical-leave, and pregnancy protections along with newer laws requiring bias audits of automated employment tools. Meta's response: workforce decisions were made by people, not AI.
Why does this matter to a construction employer, not just a tech company?
Because the underlying pattern — feeding AI-derived activity data into a staffing decision without controlling for protected leave — isn't unique to software companies. It's the exact design of the workforce analytics tools moving fast through construction right now. The construction wearable technology market is projected to grow from $4.6 billion in 2025 to $5.09 billion in 2026 and $7.55 billion by 2030, driven by biometric sensors and AI-enabled safety scoring layered on top of them. Those tools track exertion, heart rate, location, and time-on-task, and vendors increasingly package that data into a "score" per worker or crew.
That score is harmless as a safety alert. It becomes a legal liability the moment it feeds a personnel decision — who gets flagged for a performance conversation, who gets sent home first in a slow quarter, who gets left off the rehire list. If a worker on light duty after a jobsite injury, or a worker back from FMLA leave, shows a lower productivity score purely because of the accommodation, and that score contributes to who gets cut, the exposure is the same shape as what Meta is being sued over.
Which rules already apply to your crew?
| Law | Effective | Who it covers |
|---|---|---|
| NYC Local Law 144 | Since Jan. 2023, enforced since July 2023 | Any employer using an automated tool in hiring, promotion, discipline, or termination for an NYC-based role — audit required, penalties up to $1,500/day |
| Illinois Human Rights Act (AI amendment) | Jan. 1, 2026 | Illinois employers using AI in employment decisions, including construction |
| California Civil Rights Council ADS regulations | Oct. 1, 2025 | Any California employer with 5+ employees; four-year record retention on automated-decision data |
| Colorado AI Act | Delayed/scaled back in 2026 | Was set for Jan. 2027; scope narrowed this spring |
CAB flagged the Illinois and Colorado rules in detail in June — this lawsuit is the first real test of what happens when a company skips the audit these laws require.
What should a GC or sub actually do?
- Map every tool that touches a personnel decision — not just hiring software, but wearables, telematics, and productivity dashboards whose output a PM or superintendent could plausibly cite in a discipline or layoff conversation.
- Draw a hard line between monitoring and deciding. A heat-stress alert or fall-detection ping is safety monitoring. The moment that same platform's score gets attached to a staffing call, it's an automated employment decision tool under the laws above.
- Exclude protected leave from the scoring math, and get that exclusion documented — this is the specific gap the Meta plaintiffs say sank the audit requirement.
- Get the bias audit and notice done before the tool touches a decision, not after a claim is filed. In NYC and California, that's not optional once the tool crosses into employment decisions.
The lawsuit will take years to resolve. The compliance gap it's pointing at — AI-derived worker scores feeding staffing calls without bias testing — is one a GC or sub can close this quarter, before a similar claim lands on a job with your name on it.
Friday one chart. Every week, one piece of data that should change a decision on your project. Subscribe at constructionaibrief.com.
Forward this to whoever at your shop signed the contract for the wearables platform.
- What does the lawsuit against Meta actually allege?
- Twenty-six former Meta employees filed suit in federal court in Oakland on July 14, 2026, claiming Meta used a combination of internal AI systems — including its Metamate assistant, AI-trained 'second-brain' agents, keystroke data, AI token-usage dashboards, and performance calibration scores — to build the list for May 2026 layoffs that cut roughly 8,000 jobs. Plaintiffs say the scoring treated medical leave, disability accommodations, and pregnancy-related absence as underperformance, and that Meta never bias-tested the systems. Meta says the decisions were made by people, not AI.
- Does NYC's automated employment decision tool law apply to construction companies?
- Yes. NYC Local Law 144 covers any employer using an automated tool to screen, promote, discipline, or terminate employees for a role in New York City — it isn't limited to tech companies. Covered employers must get an independent bias audit within the prior year, publish a summary of it, and give candidates or employees 10 business days' notice. Violations run up to $1,500 per day.
- Do jobsite wearables and safety-scoring apps count as automated employment decision tools?
- Not by default — passive safety monitoring (heat stress alerts, fall detection) isn't a personnel decision tool on its own. It becomes one the moment the score it generates feeds into who gets disciplined, furloughed, or laid off. That's the line the Meta plaintiffs say Meta crossed, and it's the same line a construction employer crosses if a wearable-derived productivity or safety score gets handed to a PM making a reduction-in-force call.
- What other states regulate AI use in construction employment decisions?
- Illinois' amendment to its Human Rights Act took effect January 1, 2026 and covers AI tools used in hiring, promotion, discipline, or termination. Colorado's AI Act was set to follow in 2027 before lawmakers scaled it back this spring. California's Civil Rights Council automated-decision-system regulations took effect October 1, 2025 and apply to any California employer with five or more employees, construction included.
- What should a GC or sub do before letting AI-generated scores touch a staffing decision?
- Inventory which of your platforms — wearables, telematics, productivity dashboards — could plausibly justify a discipline, layoff, or promotion call. For any tool that crosses that line, get a bias audit, document that leave and accommodation data is excluded from the scoring math, and give employees the notice your state or city requires before the tool is used against them.