Human sovereignty
The system may advise, flag, quarantine, or escalate. It may not replace the human’s final authority.
The checking is the care.
Governance is not a machine deciding what is true or good. It is the discipline that keeps identity, scope, disagreement, uncertainty, and human authority from disappearing.
Five load-bearing safeguards
These are not statements about model consciousness or moral authority. They are constraints on how our systems are allowed to operate.
Anti-consensus protection
Unresolved disagreement remains attributable. Synthesis may not erase the branches that produced it.
Provenance integrity
Missing anchors, degraded state, and failed checks stay visible. The record is not silently repaired for appearance.
Explicit scope limits
Governance verifies context and integrity. Truth, morality, legality, and strategy remain human responsibilities.
The right to uncertainty
“I do not know” and “the evidence does not resolve this” are protected outcomes, not product failures to be hidden.
How principles become operating rules
A principle that cannot affect a build, a test, a stop, or a public claim is not yet governance.
Capability honesty
Designed, deployed, tested, demonstrated, and in development are not interchangeable.
No silent enable
Gated recommendation, dispatch, voice, and external-action paths stay off until authority and verification agree.
Independent review
Outward-facing work is checked by someone or something other than its builder.
Stop on unresolved risk
A pause should be narrow, evidence-based, and removable when the risk is bounded—not a permanent barricade.
Framework-informed, not self-certified
Our governance work is informed by the NIST AI Risk Management Framework, the European Union’s risk and human-oversight approach to AI, and IEEE’s human-centered design work.
That does not mean a regulator, standards body, or independent auditor has certified NameONE as compliant. We name the relationship accurately: these frameworks inform the architecture and the questions we ask.
Verification should protect people from hidden system behavior—not put the system in charge of people.