AI Security Insights

AI Security Insights

---

Reading Path

AI security is not only a model problem. In production, risk accumulates around data flows, retrieval pipelines, agent tools, memory, logging, identity, authorization, and the organizational pressure to deploy faster than governance can adapt.

Use this page as a guided path through the AI security material on this site.

Start Here

1. When the Application Fights Back: AI Security Through the Lens of Classical AppSec
   A practical bridge between traditional application security and AI-specific failure modes.

2. Prominent AI Security Frameworks: A Practical Guide for 2026
   How NIST AI RMF, MITRE ATLAS, OWASP LLM, ISO/IEC 42001, and related frameworks fit together.

3. MITRE ATLAS Deep Dive: Threat Intelligence for AI Systems in 2026
   A threat-intelligence lens for adversarial AI behavior and real-world attack chains.

LLM and Agent Security

• OWASP Top 10 for LLM Applications: An In-Depth Guide for 2026
• Prompt Injection Threat Model: A Practical Guide Using MITRE ATLAS
• LLM Agents in Production: Workflows, Frameworks, Security and Deployment

Governance and Deployment

• The Deployment Dilemma: Navigating the Challenges of AI in Regulated Environments
• Shadow AI: The Invisible Threat Inside Your Organisation
• The Architecture of Intelligence: A Deep Dive into AI Computing Infrastructure

Consulting Relevance

If you are moving an AI system from prototype to regulated production, the useful outputs are usually concrete: a trust-boundary diagram, threat model, control roadmap, logging plan, policy mapping, and architecture changes that reduce blast radius.

AI Security Architecture services describe how I approach that work.