Every CIRIS decision produces a cryptographically-signed trace. Select a trace below to see exactly how the agent reasoned.
Live data from CIRISLens • Public sample traces only
The conscience component runs checks on every action before execution.
IDMA measures information source diversity to detect fragile reasoning.
These traces aren't just logs—they're the foundation of a new approach to AI alignment. As traces accumulate, they form a corpus of validated ethical reasoning. Truth becomes simple to verify. Coordinated deception becomes increasingly constrained.
Learn why behavioral monitoring beats interpretability →