Gatekeeper: Deterministic AI Governance Showcase - Open Registration

Refusal is the resting state. Execution requires proof. Most AI safety tools today are educated guesses dressed up in policy language. The AI looks at what it's about to do, asks itself if it should, and gives an answer that might be different the next time you ask. Oak & Sparrow Systems Enterprise built something different. Gatekeeper sits between an AI agent and the action it wants to take in the real world. Before anything happens, Gatekeeper checks the action against a structured map of actual laws and returns one of three answers: GREEN, YELLOW, or RED. Same input, same answer, every time. The verdict is sealed in a cryptographic record that cannot be quietly changed later. No AI does the judging. The enforcement layer is pure logic, not language models. The system refuses by…