A broad live-web scan on 12 November 2026 surfaced 61 distinct companies and worker-facing platforms advertising or facilitating paid human work tied to AI training, evaluation, RLHF, annotation, red-teaming or expert review, backed by 34 independent source hosts. There is no authoritative global registry, so this is a discovered-candidate universe, not a proven-complete one — and not every listing is salaried employment.
Each square is one company or platform, grouped by engagement type and coloured by geographic scope. Orange outline marks weaker evidence (indirect listing, planned hiring, or unstated engagement). Hover a square for details; click a group name to filter the table.
Availability and geographic eligibility change quickly; each row keeps its source link so you can judge reliability yourself. Rows tagged "weaker evidence" rest on indirect listings or planned hiring.
| Company | AI work offered | Engagement | Scope | Availability evidence | Apply | Source |
|---|
Method: live-web scan on 2026-11-12; 61 rows, one per normalized company or platform, drawn from 34 independent source hosts — no single URL backs more than 8 rows. Each row records the AI work offered, availability evidence as worded by its source, engagement type and location scope where stated, and apply/signup and source URLs. Engagement and scope groupings on this page are simplified from the sources' own wording; 11 rows state no engagement type and are shown as "unstated". Counts measure discovered candidates, not a complete global universe. Long work descriptions are truncated in the table for space.