The Algorithm of Exclusion
The Context
Twenty-six former Meta employees have filed a federal lawsuit in Oakland, California. The claim? That Meta used an AI-powered system to select workers for mass layoffs, systematically targeting those who had taken medical leave, were disabled, or were pregnant. The algorithm relied on productivity metrics and ‘AI token usage’—a measure that inherently penalises anyone whose absence was legally protected. Meta’s spokesperson has said the claims ‘lack merit,’ but the damage to trust is already done. This isn’t an isolated glitch. It’s a pattern.
The Risk
For directors, this case is a flashing red light. Meta is already under a court-enforced DOJ settlement for discriminatory ad targeting, with compliance monitored through June 2026. The board knew algorithms could discriminate. They knew. Yet they allegedly allowed a similar logic into HR decisions. If discovery shows board-level approval of this layoff tool—or even passive awareness—directors face personal exposure under the duty of oversight. In New Zealand, the Health and Safety at Work Act 2015 and the Privacy Act 2020 impose parallel duties: you must ensure your systems don’t cause psychosocial harm or misuse health data. The ‘Court of Public Opinion’ moves fast. Brand erosion is not theoretical. It’s a balance sheet liability.
The Control
This is not a technology problem. It is a governance failure. The board must demand a human-rights impact assessment of every algorithmic tool used in employment decisions. Before the next layoff. Before the next lawsuit. You need independent audit trails that prove your AI does not discriminate against protected classes—and you need them now. Silence is complicity.
The Challenge
These are the critical questions you should be raising at the board table:
| What specific metrics does our AI-driven HR system use to rank or select employees, and have we tested them for adverse impact on disabled, pregnant, or medically absent workers? | |
| When did the board last receive a formal report on algorithmic bias risk in our employment tools, and what independent assurance did we get? | |
| If a lawsuit like Meta’s were filed against us tomorrow, what evidence would we produce to show we exercised due diligence under the Health and Safety at Work Act and Privacy Act? |