This tracks the race between AI systems becoming more capable and autonomous and our ability to monitor, interrupt, constrain, and recover from their behavior.
More autonomy does not automatically mean loss of control. The key question is whether control capacity improves at least as quickly as the pressure placed on it.
Monitoring, permissions, interruption, and recovery remain reliable as capability grows.
Control works in ordinary settings but shows brittleness under strategic or high-capability conditions.
Systems gain consequential autonomy faster than humans can reliably detect, interrupt, or redirect problematic behavior.
The state changes when admitted evidence moves the underlying indicators across the frozen scoring rules. The model should not move because a story is prominent in the news; it moves when the measured condition changes.