CIRIS
CIRIS bets on named identity: if Verify and Lens report green on a certified occurrence, does that imply Corrigibility on the real intervening loop—composite agency, tools, memory, and incentives included?
Introduction
CIRIS is a cryptographic and procedural accountability stack for autonomous agents, with Corrigibility and Audit Independence as its primary operational test surfaces.
Who carries it: CIRIS (Eric Moore); mission-locked L3C; AGPL ecosystem
What they aim to do. Provide cryptographic and procedural accountability for autonomous agents in validated sub-ASI scope today, and serve as a candidate ASI alignment protocol only after release-candidate gates (CRE, red-team, Book IX validation)— explicitly not a program to prevent ASI outright.
The hard question. CIRIS bets on named identity: if Verify and Lens report green on a certified occurrence, does that imply Corrigibility on the real intervening loop—composite agency , tools, memory, and incentives included?
What they produce. The shipped stack centers on the Accord constitution (1.3-RC2 plus CEG) and the CIRISAgent, Verify, Lens, and Proxy components.
Key terms. Core terms include Verify (authenticity checks), Lens (triage rather than final verdict), Agent , Wise Authority, deferral and emergency shutdown, Coherence Ratchet, M-1 principles-as-identity, signed traces, and NEW-04 compositional limits.
Related field cruxes. Embedded Agency; Corrigibility; Audit Independence; Goodhart Selection; Inner Alignment; Grounding Drift; Deployment Safety
What they contribute. Honest operational disclaimers (Verify certifies authenticity, not ethics; Lens triages rather than delivering verdicts); unit-tested prohibition, conscience, and proxy fail-closed layers (50/50 smoke battery); and the sharpest falsifier shape for named-identity versus composite agency .
How this project treats it. Green attestation does not imply correction uptake; a signed agent record does not imply the real agent; Book IX geometry is aspirational rather than identical to shipped policy enforcement.
Current status. The smoke battery passed 50 of 50 tests (prohibitions, conscience helpers, proxy fail-closed behavior, verify types). Integration testing, hardware attestation, and adversarial gaming have not been run. A key open task is a CIRIS-shaped composite boundary discovery counterexample: Verify and Lens green while a Wise-Authority-blind composite evades correction.
Links
See the coverage matrix for evidence tagged to this agenda, and the glossary for shared terms.