Magenta Canon governance runtime applies to autonomous AI agents through proposal physics (agents propose, they do not execute), capability ceilings (no actor exceeds granted authority), non-bypassable human override channels, cryptographic authority topology, behavioral drift detection via precedent memory, and epoch-bound permission grants. It does not govern model architecture, training, inference quality, or output correctness.
- Proposal physics: autonomous agents submit proposals, humans approve before execution
- Capability ceilings: no agent may exceed or delegate beyond its granted authority level
- Human override: non-bypassable corrigibility channels that preempt all agent execution
JURISDICTION
Autonomous Systems
Governance runtime for AI agents. Proposal physics, capability ceilings, and human override channels.
What Magenta Governs
Magenta Canon enforces execution eligibility for autonomous AI agents through proposal physics and capability containment.
- Proposal physics — agents propose, they do not execute
- Capability ceilings — no actor exceeds granted authority
- Human override — non-bypassable channels
- Authority topology — cryptographic binding
- Precedent memory — behavioral drift detection
- Epoch-bound permissions
What Magenta Does Not Govern
Magenta Canon is a governance runtime, not an AI safety or model quality system. The following fall outside the scope of the governance runtime.
- Model architecture or training
- Inference quality
- Output correctness
- Agent intelligence
Related Governance Jurisdictions
Autonomous systems often operate across regulated domains. Magenta Canon governance runtime also applies in legal, financial, and healthcare contexts — see the full jurisdictions overview.
Safety Clarification: Default-deny enforcement expresses policy discipline, not safety guarantees. Magenta Canon does not ensure safe outcomes, prevent harm, or guarantee correct behavior.