AI Layoffs Index: source-cited register of AI-attributed job cuts

Correction (2026-09-28). In the first deposit (2026-09-14) the Counts in US index column repeated whether an event’s roles were US-based, so 19 rows that count zero were marked yes, among them UPS, Verizon and Walmart, which the register declines to count. This version marks only the 2 rows whose counted figure enters the US index sum, as the data dictionary always defined it, and adds a Why not counted column that states the grounds for each of the 26 reported figures the register sets aside. Use this version rather than the first. A continuously maintained, source-cited register of workforce reductions that employers, their filings, or credible reporting have connected to artificial intelligence, with each row classified by how strongly the employer itself attributes the cut to AI, plus the transparent 0 to 100 index built on top of it. Every event links to a primary source (an SEC filing, a company statement, or reputable reporting), so each figure is independently verifiable. This deposit is a dated snapshot (methodology v1.28, snapshot 2026-09-28) of a dataset that is updated live at ailayoffs.org. For the always-current data, use the free, keyless JSON API below. This snapshot: 111 source-cited events from 92 companies, covering 2023-05-01 to 2026-09-01. 100% of events carry a primary-source URL (111 of 111), from 57 distinct source domains. 47 events are Tier 1 (a company filing or a direct company statement). 12 countries represented; 2 events enter the US index sum, and 26 reported figures are set aside with their grounds stated. Files: ai-layoffs-register.csv (the register, one row per event, 16 columns, UTF-8), README.md (full data dictionary and method summary), and LICENSE.txt (full terms). The index. The register is the evidentiary base for the AI Layoffs Index, a composite score from 0 to 100 measured relative to AI's own history since 2023: a logarithmic opinion pool of three weighted, z-scored pillars (Direct 45%, Exposure 35%, Macro 20%) mapped through a fixed logistic. At this snapshot it read 68.1 (Elevated) as of 2026-09-28; it recomputes live. Full method: ailayoffs.org/methodology. Live, always-current version (free, keyless). JSON (whole index and register): GET https://ailayoffs.org/api/index (CORS-enabled). CSV (register only): GET https://ailayoffs.org/api/register.csv. Human landing page: ailayoffs.org/data. License. Free to read, quote, redistribute, and build on, including commercially, with attribution: “Derived from free and open sources; attribute to the AI Layoffs Index and the underlying sources.” Attribute the AI Layoffs Index by name, and for any specific figure also credit that row's linked primary source. This is a plain attribution license over the project's own work, deliberately not a blanket Creative Commons grant (the register links to primary sources under mixed licenses, so a single CC grant would claim rights the project does not hold). Full terms: ailayoffs.org/license. How to cite. AI Layoffs Index: <value> (<band>) as of <date>. AI Layoffs, methodology v1.28. https://ailayoffs.org No warranty. A research observatory; every figure traces to a public source and is provided as-is. It measures AI's estimated impact on jobs relative to AI's own history since 2023, not total unemployment.