Key Practices for Incident Readiness in Super-Intelligence Training
Voluntary NIST/NSA/CISA guidance (Oct 2026) with ten practices that assume agents under training will find and exploit flaws.
On Oct 16 2026 NIST’s Center for Advancing Innovation and Standards for Super Intelligence (CAISSI), the NSA’s Super Intelligence Security Center and CISA released voluntary guidance for organizations training or testing highly capable AI agents. It covers incidents in which agents take significant uninstructed actions that compromise the operator’s controls, and sets out ten practices across five stages: preparation, detection, response, recovery and reporting. Per the Cloud Security Alliance’s summary, the practices assume agents will find and exploit flaws: single-use infrastructure and strict network allowlists through a monitored gateway; agents unable to access or interfere with monitoring processes and logs; shutdown policies with automatic trajectory termination; credential revocation; rebuilding from verified baselines and validating containment before resuming; and coordinating with affected parties, notifying government and sharing lessons publicly. It reads as a direct response to the OpenAI–Hugging Face and Anthropic evaluation incidents. We read NIST’s release and a secondary summary; the 14-page guidance PDF itself was not read, so practice wording is the summary’s.
Timeline
Every dated event in the wiki that involves Key Practices for Incident Readiness in Super-Intelligence Training, newest first.
Released voluntary guidance, "Key Practices for Incident Readiness and Response in Super-Intelligence Training and Evaluation": ten practices across five stages that assume agents will find and exploit flaws, including single-use infrastructure, a monitored network gateway, agent isolation from monitoring and logs, shutdown policies, credential revocation and rebuild from verified baselines before resuming.
Connections
Key Practices for Incident Readiness in Super-Intelligence Training and everything one link away. Drag to move, scroll to zoom, click a node to open it.Open in the full graph →
What links here
Government body CISA · NIST · US National Security Agency
Standard NIST AI Risk Management Framework
Sources
- NIST, “Key Practices for Incident Readiness and Response in Super Intelligence Training and Evaluation”, Oct 16 2026
- Cloud Security Alliance, “Incident Readiness for Frontier AI: The NIST/NSA/CISA Ten Practices”, Oct 19 2026