We are security researchers focused on AI security.
Bridgekeeper is built by Christopher Harrison, PhD and Michael Coen, PhD — researchers and engineers who have spent their careers building large-scale machine-learning systems and securing AI in regulated environments. We package that work into an in-process containment layer that reduces and contains prompt injection and data exfiltration on every LLM call, across LiteLLM, Ollama, and Bifrost.
Christopher Harrison, PhD
Founder & ML Architect
Founder of PrismScope.AI. PhD in Computer Science (Universidade do Porto) and a BS in Computer Science & Economics (University of Wisconsin–Madison). Two decades architecting high-performance and ML systems — including a 240-node HPC cluster supporting 1,500+ researchers and HIPAA-compliant ML pipelines in healthcare — with deep work in AI governance, MLOps, and serving as a CISO for AI systems. Founder and chair of the MadAI meetup and founding chair of the IEEE Computer Society's Madison chapter.
Michael Coen, PhD
Founder & Researcher
Earned his S.B., S.M., and Ph.D. at MIT, where his dissertation received the Sprowls Award for outstanding dissertation in computer science. Former faculty in Computer Science and Biostatistics at the University of Wisconsin–Madison, and co-founder of several startups. His current research focuses on hardening AI systems and LLM infrastructure — including GuardLLM, the application-layer security middleware Bridgekeeper builds on.
Maintained protection, continuously updated
New prompt-injection and exfiltration techniques surface constantly, so the protection you run has to keep moving with them. Our license-keyed protection feed — updated detection, rules, threat intelligence, and engine hardening — is maintained continuously as the attack landscape shifts. That maintained feed is the core of what we deliver.