Advanced control principles for analysis, mitigation and restoration (required, with baseline and intermediate, of an AI whose inherent risk is high). Forensic records are used, where available, to support investigation and mitigation and to improve controls; incident management works with threat intelligence during an incident, and links between intelligence, network operations and incident response allow proactive response; resilience testing is comprehensive and coordinated across all critical functions; the institution validates it can recover from events like known sophisticated attacks elsewhere and can shift processes between processing centres or systems without interruption or loss of data; testing examines from an attacker's view how assets at critical third parties could be targeted; root causes of problems found in resilience testing are corrected; scenarios with significant financial loss are used to stress-test cyber risk management. In 2.0 restoration follows a prioritised strategy with documented, timestamped actions and a check that restored systems are free of the original vulnerabilities.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.