Methodology
cutby.ai tracks job cuts that have been publicly attributed to AI, verifies each against primary sources, and labels how direct that attribution was. This page explains exactly what we count, how we classify it, and how we handle corrections. The full dataset is open.
What we track
A tracked event is a specific, announced reduction in a company's workforce where AI is named as a reason, either by the company or by credible reporting. We record the date, company, industry, market, number of jobs, attribution label, and at least one source per event. We do not track hiring freezes, natural attrition, or general layoffs with no stated AI link.
Attribution labels
Not every "AI layoff" is the same, so every event carries one label. Explicit: the company itself named AI as the reason. Mixed: the company cited AI among other reasons, such as cost or restructuring. Reported: credible reporting attributes the cut to AI, but the company did not. Walk-back: the company later softened or retracted its AI attribution; the event stays in the ledger with this label, because the reversal is part of the story.
Source requirements
Every event links to at least one public source, and we prefer primary ones: company statements, regulatory filings, WARN notices, and the monthly Challenger, Gray & Christmas report. Where figures differ between sources, we use the most-cited number and note the discrepancy on the event.
Numbers we cannot fully verify
When a company announces only a percentage or does not disclose an absolute headcount, the event is recorded with the number undisclosed and is excluded from per-event averages, so it cannot distort them. Phased or future-dated reductions are counted at announcement, which is how the industry reports them, with a note.
Cut Risk
The Cut Risk score (0 to 100) for each occupation reflects verified, AI-attributed losses weighted by workforce exposure. It is a measure of what has been tracked, not a prediction that every job in a role will disappear.
Corrections
We correct the ledger when better information appears, and because every change is a versioned edit to the open dataset, the update history is public. To flag an error or submit an event, contact us.
The data is open
The full dataset is available as CSV and JSON under CC BY 4.0: cite it, embed it, build on it, with attribution and a link to cutby.ai.