AI safety in the 2021 report
Evidence you can use
AI safety in the 2021 report
Historical snapshot: October 2021. Dates and populations are specified per row.
| Measure | Reported value | Definition and source |
|---|---|---|
| Long-term alignment researchers | Fewer than 100 | Estimate across seven selected leading organizations.2021 report, slide 157 (PDF page 157) |
| Organizations in the staffing comparison | 7 | Selected organizations, not the full AI-safety ecosystem.2021 report, slide 157 (PDF page 157) |
| Best-model TruthfulQA result | 58% truthful | Targeted benchmark result reported alongside a 94% human baseline.2021 report, slide 44 (PDF page 44) |
The staffing estimate has a narrow organizational and research scope. TruthfulQA measures a particular failure mode; it is not a comprehensive measure of alignment or all model behavior.
Sources and dates
Historical snapshot published October 12, 2021. This web edition was prepared on 2026-10-11 from the online deck and original launch posts. Findings and forecasts retain their original time frame.
- 2021 report, slide 44 (PDF page 44)Original 2021 report. Printed slide numbers match PDF pages.
- 2021 report, slide 157 (PDF page 157)Original 2021 report. Printed slide numbers match PDF pages.
- State of AI Report 2021: online slides
- Air Street Press launch essayOctober 12, 2021.
- Welcome to State of AI Report 2021Original website launch post, October 12, 2021.
Cite this page
Benaich, Nathan, and Ian Hogarth. “AI safety in the 2021 report.” State of AI Report 2021. Historical report snapshot; web edition prepared 2026-10-11.