CARICivilizational AI Risk Index
CARI reading · 23 Oct 2026
47.0 / 100
Contested control
Above appetite
scorecard base 44.5 + news adjustment +2.52
vs previous reading−0.2
vs 7 days earlier−0.4
vs 30 days earlier–
appetite 40tolerance 60capacity 80Skynet 90

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

Summary

CARI edged down 0.17 to 47.01, but every dimension remains above appetite, with misuse highest at 50.18 and diffusion and control near 48.44 and 48.26. The index was pushed upward by AI self-improvement and medical-diagnostic capability gains, agent credential deployment, and misuse involving guardrail-evading hackers and the ConnectWise breach; a smaller, faster clinical decision model lowered diffusion risk and was the only reported event pulling the index back.

Bottom line. AI risk is heading toward broader autonomous capability and operational misuse rather than retreating, with confirmed agentic attacks, recursive self-improvement, and large-scale credential delegation strengthening that trajectory and credible safeguards or declining agent performance refuting it.

Six dimensions

DimensionLevelBaseNewsTodayZone
Capability46.043.0+3.00+1.22Above appetite
Diffusion48.445.9+2.52−0.07Above appetite
Accident44.642.0+2.63+0.21Above appetite
Control48.345.7+2.59+0.00Above appetite
Governance45.644.3+1.28+0.11Above appetite
Misuse50.247.2+3.00+0.77Above appetite

Source: daily briefing email.

Change from the previous reading

40455055Capability46.0Diffusion48.4Accident44.6Control48.3Governance45.6Misuse50.2
23 Oct 2026 Previous reading (20 Oct 2026)

Stories that moved the index

Showing the largest movers of 12 stories in that day's briefing.