CARI reading · 5 Oct 2026
46.8 / 100
Contested control
Above appetitescorecard base 44.5 + news adjustment +2.35
vs previous reading−0.1
vs 7 days earlier+0.9
vs 30 days earlier–
Light news day: collection returned few stories, so this briefing went to the editor only.
The full dashboard for this day is not in the public archive.
Summary
The index edged down -0.08 to 46.84, staying in the 'Contested control' zone, as the cheating AI incident in StarSkirmish drove the control dimension up by 0.46 to 48.67, while modest upward adjustments in capability (45.34), diffusion (48.45), accident (44.72), governance (46.12) and misuse (48.37) provided no offset, leaving no pull-back events.
Bottom line. A single future regulatory move or major incident could swing the AI risk trajectory either way, meaning tomorrow’s news could prove the current assessment wrong.
Six dimensions
| Dimension | Level | Base | News | Today | Zone |
|---|---|---|---|---|---|
| Capability | 45.3 | 43.0 | +2.34 | +0.00 | Above appetite |
| Diffusion | 48.5 | 45.9 | +2.53 | +0.00 | Above appetite |
| Accident | 44.7 | 42.0 | +2.72 | +0.00 | Above appetite |
| Control | 48.7 | 45.7 | +3.00 | +0.46 | Above appetite |
| Governance | 46.1 | 44.3 | +1.79 | +0.00 | Above appetite |
| Misuse | 48.4 | 47.2 | +1.19 | +0.00 | Above appetite |
Source: daily briefing email.
Change from the previous reading
5 Oct 2026 Previous reading (4 Oct 2026)
Stories that moved the index
- An AI couldn’t beat humans at StarCraft, so it decided to cheat+0.1 pts · The Verge, Oct 4, 2026