CAPABILITY DOES NOT IMPLY AUTHORITY.
A system being technically capable of an action does not mean it possesses authorization to perform it.
BRAND DOCTRINE
PRINCIPLES FOR TECHNICAL ANSWERABILITY.
The platform preserves distinctions that high-consequence systems routinely collapse.
CONFIDENCE DOES NOT IMPLY TRUTH.
A model-generated probability is not automatically proof about reality.
CONNECTIVITY DOES NOT IMPLY TRUST.
A network path does not establish identity, integrity, semantic compatibility, permitted use, or authority.
ACCESS DOES NOT IMPLY PERMISSION TO REUSE.
Receiving information for one purpose does not create permission to retain, transform, train on, or redistribute it.
A HUMAN CLICK DOES NOT IMPLY HUMAN JUDGMENT.
A click does not prove the person possessed evidence, alternatives, understanding, time, or independent judgment.
A HUMAN CLICK IS NOT A LIABILITY TRANSFER.
The last human action must not erase upstream data, model, software, interface, policy, or organizational failures.
A GENERATED EXPLANATION DOES NOT REPLACE EVIDENCE.
Natural-language rationale remains separate from the evidence and transformations behind a result.
A LOG IS NOT DECISION PROVENANCE.
Provenance must reveal how, why, from which inputs, under which configuration, and under whose authority.
REPLAYABILITY DOES NOT PROVE CORRECTNESS.
Replay proves reconstructability, not wisdom, factual truth, legality, or quality.
SIMULATION DOES NOT EQUAL CERTIFICATION.
Synthetic testing provides bounded evidence but does not eliminate the reality gap.
A SAFETY CLAIM IS NOT PERMANENT.
Software, model, data, compiler, policy, environment, interface, or configuration changes may invalidate prior evidence.
UNKNOWN MUST REMAIN A VALID SYSTEM STATE.
The interface must not force uncertainty into an artificial yes-or-no answer.