This page may contain stale information. Last updated: 2026-07-12
Definition
The AI Safety Index is future-of-life-institute’s biannual benchmark grading leading AI companies across 37 indicators in six domains: dangerous capabilities, governance, transparency, misuse safeguards, existential safety, and military applications.
Summer 2026 Results
| Rank | Company | Grade | Score |
|---|---|---|---|
| 1 | anthropic | C+ | 2.66 |
| 2 | openai | C | 2.28 |
| 3 | google-deepmind | C | 2.01 |
| 4 | meta | D+ | — |
| 7 | xai | F | — |
No company scored above C+. Industry-wide weakest domains: existential safety, pause commitments, military AI pivots.
Related
- future-of-life-institute
- ai-safety
- ai-governance
- competitive-ai-safety-testing
- frontier-ai-governance