Technical AI security insights
Each insight solves one engineering or governance problem in depth, with the evidence behind it.
Enterprise AI defence
Technical · 10 min read
Mitigating Indirect Prompt Injection in Multi-Tenant RAG Pipelines
How untrusted documents, emails and web pages hijack RAG sessions, and the privilege boundaries and deterministic output checks that contain the damage.
Technical · 11 min read
Securing Autonomous AI Agents: Deterministic Gateways for Tool Use and Function Calling
AI agent security fails when agents hold their own credentials. A policy gateway, strict tool-call schemas and rare, clear approvals contain excessive agency.
Technical · 11 min read
Row-Level Security and Tenant Isolation in Vector Databases
Similarity search can leak one tenant's data to another. Enforce namespace isolation and tenant filters before retrieval, not after the model answers.
Technical · 11 min read
Red-Teaming Enterprise AI: An Architectural Audit Checklist for the OWASP LLM Top 10
An AI red teaming checklist for the OWASP LLM Top 10: what to test, why static prompt lists mislead, and the evidence every finding needs to stand up.
