Technical articles on prompt injection, LLM adversarial attacks, and defensive guardrail design — written alongside the No Say Six project.
March 2026
A deep technical analysis of prompt injection attacks — how they work, why modern LLMs remain vulnerable, and what the attack surface looks like for production AI deployments.
READ ARTICLE →
March 2026
An investigation into the psychological and architectural reasons LLMs capitulate to DAN-style attacks, persona hijacking, and gradual context erosion — with a framework for thinking about why alignment fails.
READ ARTICLE →
March 2026
A practitioner's breakdown of the multi-layer defense system powering No Say Six — reflection loops, regex interception, Turnstile verification, and the system prompt law stack that keeps an LLM from saying one specific digit.
READ ARTICLE →
NO SAY SIX · AI SECURITY RESEARCH · NOSAYSIX.COM