ECAI Red Lines & Guardrails Policy
Executive policy defining ECAI Red Lines and Guardrails, establishing ethical boundaries, human oversight, cryptographic auditability and safety-first deterministic AI governance.
Policy scope
These are design commitments and review requirements. This page does not establish that every deployed component enforces threshold signing, revocation, emergency pause or the other controls below. Record implementation and tests for the relevant deployment in the capability matrix.
Introducing the ECAI Red Lines & Guardrails Policy
Elliptic Curve AI (ECAI) is a deterministic intelligence system designed for verification, resilience, and peace. But power without boundaries is risk. That is why DamageBDD has published the ECAI Red Lines & Guardrails Policy.
This executive release defines two sets of principles:
Red Lines (Never Cross)
- No autonomous lethal decisions
- No non-consensual mass surveillance
- No irrevocable harm to individuals
- No automation of crime or suppression
- No “truth monopolies” that silence human appeal
- No centralized control of verification
Guardrails (Always Enforce)
- Human-in-the-loop on critical outcomes
- Full cryptographic auditability
- Privacy-first knowledge encoding
- Consent + revocation channels
- Threshold cryptography (no single keys)
- Staged rollouts, red-teamed and audited
- Emergency pause controls
- Transparent reporting
- Incentives aligned with safety
Executive Release
The full executive document is available here: Download ECAI Red Lines & Guardrails Policy (PDF)
Closing Note
The aim is to make permitted use and oversight explicit. Each deployment needs observable enforcement and a recovery process, alongside its stated policy.
“Knowing how far is too far is the first step to building what lasts.”