Researcher uses differential geometry and Fisher information theory to build Arc Gate, an LLM security proxy that detects prompt injection attacks with measurable performance metrics.

Arc Gate is a monitoring proxy that sits between applications and LLM APIs (OpenAI, Anthropic) with only a single URL configuration change required

The tool is based on five published research papers on second-order Fisher information manifold (H² × H², R = −4) that identifies a phase transition threshold at τ* ≈ 1.2247

Both DistilBERT and GPT-2 XL models converge to the theoretical threshold τ ≈ τ* during training, validating the mathematical approach

Arc Gate provides real-time behavioral monitoring, injection blocking, and a dashboard with transparent performance metrics unlike most vague AI security tools