What operating modes does AI Control support?

Five: HUMAN ONLY (AI operations disabled), AI RESTRICTED (limited, low-sensitivity capabilities only), AI SUPERVISED (sensitive actions require human approval), CONTROLLED AUTONOMY (approved actions run automatically within strict policy) and NEUTRALIZE (defensive containment and decommissioning of administered infrastructure). The decision engine enforces the mode on every AI action; changing it is a human-only, audited governance action anchored in the Solieum Governance Registry.

Frequently asked questions

Does HUMAN ONLY stop local safety controllers?

No. Modes govern AI authority. Local safety controllers, emergency stops and safe states are independent and unaffected.