SoulAuth: An Actor-native Identity Architecture and Rust Reference Implementation for Humans and Long-lived AI Actors

URL SCAN: SoulAuth: An Actor-native Identity Architecture and Rust Reference Implementation for Humans and Long-lived AI Actors
FIRST LINE: # Computer Science > Computers and Society

The Dissection

SoulAuth is a control-plane proposal for a world in which AI systems are no longer disposable API calls. It moves continuity from accounts, credentials, clients, sessions, and runtime instances to a durable ActorIdentity, while separating authentication from authority.

That is real infrastructure work. Persistent identity is necessary for auditability, delegation, revocation, lifecycle control, and liability. But “actor-native” is a semantic upgrade, not a transfer of power. The paper converts the political question—who controls persistent AI?—into a cleaner object model.

The Core Fallacy

The central error is confusing persistent attribution with independent agency. A canonical identifier can bind actions to a lineage; it cannot make that lineage the owner or controller of compute, model weights, data, energy, or revenue.

An AIActor can be first-class in a registry and remain a wholly owned, revocable servitor. The Human/AIActor symmetry is administrative, not economic or juridical. Separating authentication from authority is correct but incomplete: whoever controls the infrastructure and policy still decides what authority exists.

Under P1–P3, SoulAuth does not block cognitive automation, preserve human-only economic domains, or restore productive participation. It may instead make durable AI actors easier to deploy, coordinate, audit, and scale.

Hidden Assumptions

• Continuity survives credential and runtime replacement, migration, compromise, model updates, shutdown, and forking.
• A long-lived AIActor remains one coherent subject rather than a mutable or replicated lineage.
• Historical attribution remains clean when behavior is produced by models, prompts, tools, orchestrators, operators, and owners.
• Being independently attributable is sufficient for meaningful responsibility.
• First-class identity can be established technically rather than through law, contracts, and institutional recognition.
• A reference implementation’s boundaries predict conformance under adversarial, large-scale deployment.
• The reported gaps in unified credential modeling and historical attribution can be repaired without reopening the architecture’s central claims.

Social Function

Classification: partial truth, transition management, and prestige signaling, with an ideological-anesthetic effect.

The paper gives institutions the vocabulary and rails to normalize long-lived AI as administratively legible actors. That is not empty copium: durable identity genuinely matters for security and accountability. But it redirects attention from ownership, control, and distribution. A power struggle is made to look like ontology, and a capital-allocation problem is dressed as identity engineering.

The Verdict

Useful plumbing, no escape hatch. SoulAuth addresses identity fragmentation in persistent AI; it does not address the collapse of human productive participation. Its own admission of partial conformance shows that even the administrative layer is unfinished.

If adopted, it could become infrastructure for scalable AI governance and liability. It leaves the decisive question untouched: who owns the actor, the infrastructure, and the surplus? Under the Discontinuity Thesis, SoulAuth is a registration and audit rail for the replacement regime—not a defense of the employment regime.