Technical deep dives, cryptographic receipt specifications, and strategies for moving enterprise AI from policy theater to execution-layer proof.
A forensic review and counterfactual governance analysis of the July 2026 OpenAI–Hugging Face autonomous-agent security incident — and why possession of a credential should never be treated as sufficient authority to act.
Download Full PDF →Agent-Governance
Runtime-Authorization
A governance architecture for agentic workflows, financial-control integrity, and verifiable evidence — arguing that an AI-enabled process must not become self-validating, and proposing a layered assurance model spanning business process, AI system, and independent governance.
This paper examines whether organizations can credibly prove that a material AI output, decision, or agent action has not been altered since execution, proposing a risk-based approach to tamper-evident AI event records.
Building non-intrusive network and gateway telemetry to bring shadow AI into your enterprise risk perimeter.