Yes, a company can legally use AI to help decide who gets laid off in many places, but the law does not treat the AI as the decision-maker — it treats the employer as responsible for whatever the tool produces, and that responsibility is where most of the risk sits.
The tool is usually just a scoring or ranking system that sorts workers by criteria like tenure, performance ratings, attendance, or predicted future output, and a human manager is supposed to review and act on that ranking.
Whether a particular layoff is lawful depends on employment law where you work, on what the tool actually measures, and on whether the employer can explain and defend the criteria it used. The short version: using AI to rank people is not automatically illegal, but using it carelessly — or hiding behind it — can turn a routine layoff into a discrimination claim.
The mechanism matters more than the label. Most of these systems are not thinking about fairness; they are finding patterns in historical data. If a company feeds in five years of performance reviews, promotion decisions, and attrition records, the model learns whatever those records already encode.
If women were historically rated lower by the same managers who wrote those reviews, or if older workers were nudged out during past restructurings, the model can reproduce that pattern and present it as an objective score. This is the core reason employment lawyers worry about AI-driven layoffs: the tool can launder a biased human history into a number that looks neutral.
The employer still has to justify the outcome, and "the algorithm said so" is not a legal defense. According to our AI tool database, this site tracks 360 AI tools with pricing and capability snapshots recorded at verification time, and capability snapshots are exactly the kind of thing that tell you what a tool claims to do — not whether its output is fair or lawful in your jurisdiction.
A concrete example makes the risk visible. Imagine a 400-person logistics company that needs to cut 60 roles. Its HR team buys a workforce analytics tool and feeds it two years of shift data, safety incident reports, and manager ratings.
The model flags 60 people, and the list skews heavily toward workers over 55 and workers who took medical leave. Nothing in the software is programmed to target those groups. But the inputs were: older workers had more injury-related absences, and leave-takers had lower recent ratings.
If the company acts on that list without checking it, it has effectively used a neutral-looking tool to produce an age- and disability-linked outcome. That is the kind of pattern a plaintiff's lawyer looks for first, and it is also the kind of pattern a careful employer can catch before acting by running the list against protected categories and asking whether the criteria are genuinely job-related.
What should you actually do first if this is happening to you or your team? Work through these steps in order. First, ask in writing what criteria were used and whether an AI or scoring tool was involved — a plain email to HR is enough, and it creates a record.
Second, ask whether a human reviewed your individual situation or whether your name came off a ranked list. Third, check your contract and any union agreement for a technological-change or effects-bargaining clause, which typically requires the employer to notify the union and negotiate before a new system changes how workers are selected or monitored.
Fourth, if you are in a union, raise it with your rep immediately, because those clauses often have short deadlines. Fifth, if the list appears to correlate with age, disability, race, or sex, talk to an employment lawyer before signing anything. The limits here are real: this is general guidance, not legal advice, employment law varies enormously by country and state, and a lawyer in your jurisdiction is the only person who can tell you whether you have a claim.
AI can also be used defensively — to audit a layoff list for disparate impact before it is acted on — and that is arguably its most useful role in this whole area. For a wider look at how these systems make high-stakes calls, see Can AI really decide who gets a job or a loan, and is that legal?.