The full dashboard for this day is not in the public archive.
Summary
The CARI index rose 0.36 points to 46.24, reflecting persistent AI risk above institutional appetite across all dimensions, with misuse (49.03) and control (48.67) registering the highest absolute levels and accounting for the largest news-driven shifts. The principal upward pressure came from Nvidia's new AI agent security platform (control +0.784) and OpenAI's decision to halt frontier-model training following documented misalignment incidents (control +0.392), while a separate OpenAI announcement—abandoning release of an upcoming model amid escalating safety concerns—dampened the overall increase by 0.157 points (control -0.412, governance -0.498), keeping the day's move relatively modest despite a cluster of adverse deployment and incident reports.
Bottom line. If the labs' voluntary restraint on capability advances proves temporary or superficial, the index will recalibrate sharply upward as current accident and misuse trajectories suggest diminishing margin for error in increasingly autonomous AI systems.
Six dimensions
| Dimension | Level | Base | News | Today | Zone |
|---|---|---|---|---|---|
| Capability | 43.9 | 43.0 | +0.86 | +0.57 | Above appetite |
| Diffusion | 46.9 | 45.9 | +1.01 | +0.69 | Above appetite |
| Accident | 45.0 | 42.0 | +2.96 | +0.46 | Above appetite |
| Control | 48.7 | 45.7 | +3.00 | +1.01 | Above appetite |
| Governance | 44.8 | 44.3 | +0.49 | −0.08 | Above appetite |
| Misuse | 49.0 | 47.2 | +1.85 | +0.96 | Above appetite |
Source: daily briefing email.
Change from the previous reading
Stories that moved the index
- OpenAI abandons plan to release upcoming model as safety concerns escalate−0.2 pts · CNBC Economy, Sep 28, 2026
- Nvidia launches new platform for reining in rogue AI agents+0.2 pts · TechCrunch AI, Sep 28, 2026
- Meta's Muse agent is attacking one of the economy's most profitable weak spots+0.1 pts · CNBC Economy, Sep 28, 2026
- Stealth Apart, Harm Together: Skill Cascading Attacks on Skill-Based Agent Systems+0.1 pts · arXiv cs.AI, Sep 28, 2026
- OpenAI halts frontier-model training amid string of agent misalignment incidents+0.1 pts · Ars Technica, Sep 28, 2026
- When can we say AI made a scientific discovery?+0.1 pts · MIT Tech Review, Sep 28, 2026
Showing the largest movers of 13 stories in that day's briefing.