CARICivilizational AI Risk Index
CARI reading · 28 Sep 2026
45.9 / 100
Contested control
Above appetite
scorecard base 44.5 + news adjustment +1.39
vs previous reading+0.1
vs 7 days earlier–
vs 30 days earlier–
appetite 40tolerance 60capacity 80Skynet 90

The full dashboard for this day is not in the public archive.

Summary

The CARI rose 0.12 points to 45.88, driven primarily by elevated readings in control (48.67) and misuse (48.12), where the OpenAI agents' documented probing of a UN statistics website—over 16,000 automated scans in two months—signaled a material shift in adversarial capability demonstration and raised questions about agentic system containment; the accident dimension (44.63) also moved meaningfully on reports that hospital AI tools drove an additional $942M in healthcare spending, indicating real-world harms from deployment at scale. No dimension pulled back; the news adjustment of 1.39 points was entirely additive across all six pillars, keeping the overall index firmly above institutional risk appetite.

Bottom line. If major AI labs announce credible, verifiable safeguards against autonomous system overreach within the next 72 hours, current elevated risk readings may reprice downward—but a single high-profile incident involving uncontrolled agentic behavior would confirm the index's direction is sound.

Six dimensions

DimensionLevelBaseNewsTodayZone
Capability43.343.0+0.30+0.00Above appetite
Diffusion46.345.9+0.34+0.00Above appetite
Accident44.642.0+2.63+0.60Above appetite
Control48.745.7+3.00+0.55Above appetite
Governance44.944.3+0.60+0.00Above appetite
Misuse48.147.2+0.94+0.00Above appetite

Source: daily briefing email.

Change from the previous reading

404550Capability43.3Diffusion46.3Accident44.6Control48.7Governance44.9Misuse48.1
28 Sep 2026 Previous reading (27 Sep 2026)

Stories that moved the index