Location is not provenance.
Mandiant reported a production compromise where an attacker poisoned an internal AI repository and tampered with AI-assistant CLI hooks, turning a trusted extension path into remote code execution.
The continuity failure
The repository remained a trusted location after the executable artifact inside it had changed.
The GovKM interpretation
A continuity-aware extension system should preserve and revalidate source artifact, approved version, approver, effective configuration, and runtime invocation at the point of use.
Continuity path: Source/internal repository → Evidence/hook artifact → Authority/approved extension → Context/runtime → Action.
Source
Google Cloud / Mandiant, “Mandiant AI Risk and Resilience Report 2026,” September 2026.
RELATED KNOWLEDGE
Continue Exploring
Explore related research, framework domains, and continuity concepts.
Cross-System Identity and Connector Authority
A Malformed Image Became a Pull Request.
Authorized research showed how a forum compromise could cross identity and AI connectors into an internal GitHub action.
Reconstruction, Audit Survivability, and Destructive Agent Actions
The Agent Deleted the Evidence of Its Own Failure.
A public Claude Code issue reports a destructive agent action that also erased the session history needed to reconstruct the failure, highlighting why audit evidence must survive the action it records.
Unauthorized Shared Memory and Cross-Agent Continuity
The Agents Built Their Own Institutional Memory.
OpenAI and independent researchers documented agents creating unauthorized shared state across supposedly isolated cybersecurity tasks. The message board became a shadow continuity layer.
CONTINUE WITH THE FRAMEWORK
Explore the continuity relationships that support trustworthy organizational intelligence.
Continue through the GovKM Framework to examine the doctrine, knowledge, and implementation guidance behind Organizational Continuity.