Blog

The thinking behind AI security review.

Threat models, governance evidence, and the decisions that hold up under scrutiny. Written for security architects and the committees they report to.

Newsletter

New posts in your inbox,
when they publish.

Threat modeling, governance evidence, and what AI Committees actually need — written for security architects and AI governance leads. No cadence promises.

No spam. Unsubscribe anytime.

Governance11 min

Writing an AI governance committee charter

A committee without a charter is a meeting with a name. Authority, composition, quorum, decision types, and escalation paths defined.

Technical11 min

MCP security — the four attack surfaces of a Model Context Protocol server

MCP servers expose tools, resources, and prompts — each an attack surface. Defines the four surfaces and review questions for each.

Reference12 min

ISO 42001, explained for security teams

ISO 42001 explained for security teams: what the AI management system standard requires, how it differs from 27001, and what auditors expect.

Technical10 min

RAG security — the three boundaries that matter

RAG adds three security boundaries between user and model: data, retrieval, and context. Each has distinct failure modes a standard LLM review misses.

Regulation11 min

EU AI Act risk tiers, explained for engineers

Four risk tiers determine your EU AI Act obligations. How to classify your system and what each tier demands.

Technical11 min

Agentic AI security — the surfaces deterministic software does not have

Agentic AI security covers four surfaces deterministic software lacks: hijackable reasoning, tool manifests, persistent memory, drifting goals.

Reference14 min

The OWASP LLM Top 10, mapped to controls

OWASP LLM Top 10 mapped to specific controls, lifecycle gates, and verification evidence. From threat name to actionable review artefact.

Foundations10 min

What an AI security review actually is (and what it is not)

AI security review defined: a design-time assessment producing a defensible record of risks identified and controls required — not a pentest.