Adaptive AI security and red teaming built by practitioners with DoD roots

PurpleSec is a Washington, DC-based AI security company founded in 2019 that helps enterprises close the gap between rapid AI adoption and practical, durable security. The company blends offensive and defensive cybersecurity expertise drawn from high-consequence government environments, operating as an embedded partner rather than a tool vendor. Services span AI red teaming, consulting, policy design, and program development, with free NIST AI RMF and EU AI Act-aligned policy templates used by thousands of security teams to operationalize responsible AI.

Its flagship product PromptShield is an intent-based Prompt WAF that protects enterprise LLM applications against prompt injection, jailbreaks, data exfiltration, and obfuscation attacks across multiple model providers. The platform pairs a detection engine that analyzes intent and context with a self-training adaptive defense layer, a consistency layer to catch cross-LLM manipulation, and automated risk scoring for auditability. Red team engagements use the STRIDE-AI framework to probe novel attacks, and the whole stack is designed for ISO/IEC 42001 governance out of the box.