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.
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.
Goal hijacking and instruction drift in autonomous agents
Goal hijacking makes an agent pursue unintended objectives. Instruction drift is the slow version. Both evade detection because the agent appears to work.
Guardrails that work vs guardrails that look like they work
Most LLM guardrails are bypassable classifiers. Distinguish patterns that provide genuine risk reduction from those that only provide the appearance of it.
What evidence an AI security review should produce
Review evidence must survive regulator questions, procurement audits, and post-incident inquiries. What it needs to include.
Finding the AI vendors no one formally approved
Shadow AI is in every organization. Where unapproved AI tools hide, how to find them, and how to formalize without alienating adopting teams.
Roles and responsibilities under ISO 42001
ISO 42001 requires documented AI management roles. Who the standard expects, how they map to org structures, and what each must demonstrate.
MCP security vs traditional API security — what changes
MCP looks like a REST API but the client is a non-deterministic model. The attack surface includes tool descriptions and trust sits at the model.
The EU AI Act timeline and what to prepare first
EU AI Act provisions apply in phases. Which obligations hit when, and the preparation steps that deliver value first.
Running RAG over regulated data — the review checklist
RAG over GDPR, HIPAA, or financial data needs controls at data, retrieval, and output layers. Checklist mapped by data class with evidence requirements.
The OWASP Agentic Top 10, explained for security reviewers
OWASP Agentic Top 10 explained for practitioners: each threat category with actionable controls and evidence requirements a security reviewer can use.
System prompt leakage and why it matters for security
System prompts encode trust assumptions, scoping rules, and sometimes credentials. When they leak, the system's entire trust model is exposed.
A lightweight AI security review for fast-moving teams
Minimum-viable AI security review for fast-moving teams: three questions, three artifacts, one decision record.
What makes an AI decision record defensible
A defensible decision record lets a regulator understand what was decided and why -- without the people who made it. The required standard.
Free resources
Practical templates for every framework covered here.
AI Security Review Template
Full evidence pack with threat model, controls, and evidence grading.
OWASP Agentic Top 10 Controls
Each risk mapped to required controls and lifecycle gates.
Decision Memo
Clearance decision template with rationale and sign-off log.
AI Go-Live Security Checklist
Production gate checklist for security architects and CISOs.