Autonomic Identifier (AID)

An AID is cryptographically derived from the public key of its first inception event. This means the identifier itself is a hash of the initial public key, creating a self-referential binding between the identifier and its controlling key pair. Anyone can verify this binding mathematically without querying an external registry. This eliminates the root of trust dependency on blockchains or certificate authorities, making the identifier independently verifiable.

The core security mechanism of an AID is the Key Event Log (KEL) , an append-only, hash-chained sequence of signed events. Each event establishes the current authoritative key state and cryptographically commits to the next key hash in a pre-rotation scheme. This means:

• A compromised key cannot be used to forge a rotation because the next authorized key was already committed.
• The entire history of key state is a verifiable data structure that can be replayed to validate the current state.

Unlike Decentralized Identifiers (DIDs) that often rely on a Verifiable Data Registry (such as a distributed ledger) for state resolution, AIDs are designed to operate independently of any consensus ledger. The KEL itself is the authoritative source of truth. This provides ambient verifiability, meaning the identifier's state can be validated by any party who receives a copy of the event log, without network calls or transaction fees. This is critical for high-throughput, low-latency, and air-gapped systems.

KERI's security model is based on duplicity detection rather than global consensus. If a controller signs two conflicting versions of a key event (a fork), this act of duplicity is mathematically provable by any witness holding both event receipts. The protocol does not prevent the attack but guarantees non-repudiable detection of malfeasance. This shifts the security guarantee from Byzantine Fault Tolerance consensus to a simpler, more scalable accountability model.

To ensure availability and duplicity detection, AIDs utilize a configurable network of Witnesses and Watchers:

• Witnesses: Designated nodes that receive, acknowledge, and store copies of key events to prevent the controller from suppressing a rotation.
• Watchers: Any entity that observes the KEL and can alert the ecosystem if duplicity is detected. This creates a distributed trust layer without requiring a global consensus ledger, allowing for flexible and jurisdiction-specific deployment.

AIDs support both rotation (transferring control to a new key pair) and delegation (authorizing another AID to act on its behalf). Rotation events are recorded in the KEL, maintaining a continuous, verifiable chain of custody. Delegation is achieved through a hierarchical key event structure, enabling complex organizational identity models where a root AID can delegate specific authorities to subordinate AIDs without compromising the root of trust.