AI Midnight — 40.1/100 (Elevated)

2026-W04 · +2.1

Week of 11–18 January 2026. A cluster of real AI harms (Grok deepfakes, AI mistakes in policing and public services) and another large compute deal push the index up slightly.

• AI autonomy and loss of control : 33.5/100
• Dangerous capabilities : 39.5/100
• Real-world AI incidents : 41/100
• Governance and safety safeguards : 51.5/100
• Acceleration of technological capabilities : 42.5/100

A conventional composite index, not a scientific probability of extinction.