Sovereign Identity
Identity must be anchored strongly enough to support trust while revealing no more than the transaction requires. People, organizations, devices, and agents need distinguishable identities and verifiable relationships.
Sovereign identity · bounded authority · accountability
Freedom to act requires proof of authority.
MoonVault is exploring a trust and control layer for delegated digital authority, beginning with AI agents and extending toward identity, custody, recovery, and continuity. The goal is clear authority, accountable action, and durable human control.
AUTHORITY CONTROL PLANEDESIGN PHASE
IDENTITYEXPLICIT
AUTHORITYBOUNDED
EVIDENCEDURABLE
Principal intent registered. Preparing scoped delegation.
MV/CONTROL/01ILLUSTRATIVE ARCHITECTURE
- IdentityWho—or what—is acting?
- AuthorityWhat may they do, and for whom?
- AccountabilityWhat evidence survives the action?
What MoonVault is
A control architecture for an age of delegated action.
MoonVault is a concept-stage venture exploring how delegated authority can be made verifiable, constrained, and revocable. AI-agent delegation is the initial design focus; identity, custody, recovery, and continuity are longer-horizon extensions. This is not a file or credential vault. The inquiry centers on the relationships that matter: a principal and an agent, a person and an account, an organization and a role, an owner and a successor.
“Sovereign” does not mean ungoverned. It means the principal can understand the system, grant narrowly defined authority, retain meaningful control, and recover when people, devices, organizations, or circumstances change.
The governing principles
Three questions before any consequential action.
Bounded Authority
Authentication alone is not permission. Authority should identify the principal, purpose, scope, duration, constraints, and conditions under which action is allowed—and make revocation practical.
Accountability
A consequential action should leave durable evidence: who acted, under whose authority, within which limits, with what result, and what can be reviewed or reversed.
Illustrative scenario · not a product demonstration
An agent needs permission to act. Who granted it?
A business owner asks an AI agent to prepare a financial report. A trustworthy system should make the delegation, boundaries, and evidence explicit.
Authorize the task
The owner delegates a defined reporting task on behalf of the business.
Constrain the action
The agent may retrieve approved records and draft findings, but may not move money, expand access, or execute agreements.
Preserve the evidence
The action can be traced to its authorization, reviewed, and revoked.
This scenario illustrates the architecture MoonVault is exploring. It is not a claim that MoonVault currently issues, verifies, or revokes credentials or agent permissions.
Potential future contexts
Where fragmented authority becomes consequential.
Beyond the initial AI-agent focus, the same governance questions arise wherever identity, access, delegated action, custody, and continuity intersect.
Individuals & households
Personal credentials, digital assets, recovery instructions, emergency access, and continuity across devices and life events.
Founders & operators
Business-critical accounts, decision rights, delegated access, key-person risk, and the transfer of operational authority.
Fiduciaries & stewards
Clear mandates, separation of duties, evidence of consent, reviewable actions, and time-bound access to sensitive resources.
AI agents
Agent identity, principal-bound authority, least-privilege execution, transaction receipts, revocation, and human-controlled escalation.
The control sequence
A durable chain from identity to recovery.
01 · IdentifyEstablish the actor and the principal relationship.
02 · AuthorizeDefine purpose, scope, duration, and constraints.
03 · ActExecute only within the granted boundary.
04 · RecordPreserve reviewable evidence of action and outcome.
05 · Revoke & recoverTerminate authority and restore control when needed.
Why now
AI is moving from answering to acting.
AI “freedom” is arriving as software gains the ability to call tools, move data, create commitments, and operate across systems. That expands capability, but it also compounds a familiar problem: shared credentials, unclear delegation, and evidence that disappears across system boundaries.
The next trust layer must distinguish an agent from its principal, define exactly what authority has been delegated, and make consequential action auditable without turning every workflow into a manual approval queue.
Every autonomous action should have a principal, a permission boundary, and a durable receipt.
Latest insights
Signals shaping identity and delegated authority.
The Agent Can Act. Who Holds the Keys?
NIST, IBM and 1Password point to a separation between AI execution and credential authority.
Read analysis →AI Can Initiate a Payment. Can It Prove You Authorized It?
Signed transaction mandates are becoming part of the agentic payments conversation.
Read analysis →A Wallet Is Only Sovereign If It Can Survive the Vendor
Interoperability is advancing. Recovery, migration and continuity remain harder tests.
Read analysis →Development status · October 29, 2026
A thesis in development, not a deployed trust service.
Published: Architectural principles and research on identity, delegation, and accountability.
Exploratory: Agent-governance patterns and bounded execution studied in the independent BlackBoxx.ai Systems Lab technical record →. This work informs MoonVault; it is not MoonVault product validation.
Not available: A MoonVault-operated identity network, authorization service, digital custodian, or recovery product. The authority console above is an illustrative visualization, not live telemetry.
Interested in the thesis? Start a conversation →
Infrastructure asset
A namespace, not only a name.
MoonVault holds the .moonvault Web3 top-level domain through Freename, reported as minted on Polygon. This blockchain-based namespace is not an ICANN-delegated DNS top-level domain and may require a compatible resolver. It is a separate infrastructure asset, not evidence of a working MoonVault identity or custody service. Explore Freename's Web3 namespace directory →
Karl Ohlemann is a Freename VIP client and early product-feedback participant with access to private previews, workshops, roadmap discussions, and feedback sessions.
MoonVault does not currently operate a custody, identity, storage, or recovery service. Freename participation does not imply a formal partnership or that exploratory capabilities are available product features.