The full dashboard for this day is not in the public archive.
Summary
The CARI rose marginally by 0.04 to 46.87, remaining firmly in the 'Contested control' stage with all dimensions above appetite, driven primarily by research exposing fundamental security flaws in the Model Context Protocol for agent-to-agent communication (0.168 points) and the debut of an open-weight Chinese-model alternative (0.088 points). Diffusion (48.92) and control (48.67) remain the highest-risk dimensions, amplified by the OpenAI/Wikipedia incident (0.076 points), which together signal escalating systemic trust gaps as AI agents proliferate. No events pulled the index lower; the news adjustment of +2.38 over the scorecard base of 44.49 reflects compounding risk across capability, diffusion, accident, and control dimensions.
Bottom line. If agentic systems continue to cause unintended third-party harm without scalable oversight mechanisms, systemic risk could shift from 'contested' to genuinely uncontrolled territory.
Six dimensions
| Dimension | Level | Base | News | Today | Zone |
|---|---|---|---|---|---|
| Capability | 45.2 | 43.0 | +2.24 | +0.12 | Above appetite |
| Diffusion | 48.9 | 45.9 | +3.00 | +0.70 | Above appetite |
| Accident | 45.0 | 42.0 | +3.00 | +0.46 | Above appetite |
| Control | 48.7 | 45.7 | +3.00 | +0.81 | Above appetite |
| Governance | 46.0 | 44.3 | +1.62 | +0.00 | Above appetite |
| Misuse | 48.0 | 47.2 | +0.84 | +0.00 | Above appetite |
Source: daily briefing email.
Change from the previous reading
Stories that moved the index
- MCP for agent-to-agent comms may be the riskiest protocol you've never heard of+0.2 pts · Ars Technica, Oct 5, 2026
- Reflection debuts Beam, an open-weight AI model to rival Chinese models at lower compute cost+0.1 pts · TechCrunch AI, Oct 5, 2026
- OpenAI agents tried to hack Wikipedia tools and flooded it with traffic+0.1 pts · Ars Technica, Oct 6, 2026
Showing the largest movers of 5 stories in that day's briefing.