Earned trust and reversible delegation
Delegation that expands on demonstrated use and contracts the moment the user wants it gone. A user can withdraw delegation at any point without losing access to the outcome they came for.
The impact of agents
Agents that can hold standing delegation and act over time are now widely available. As government lets users delegate more to them, the amount of authority a user has handed over deepens. That authority can outrun or outlast the trust that justified it: an agent that keeps an expanded scope after a bad experience, or one still acting under authority the user no longer wants it to have.
What must be verified
Government needs delegation to an agent to expand only on demonstrated use and to contract the moment a user wants it to. The agent must be demoted when it crosses a boundary, and revocation must be at least as easy as granting it. The platform or service operator must enforce that demotion and guarantee revocation parity.
Protecting access
Cautious users, and those who distrust automation, are the ones most likely to need this exit. If the non-agent pathway is slower or thinner, the right to revoke becomes a penalty for revoking. A trust history used as a precondition denies service to newcomers outright.
Keeping the path open
- Start every user at the most conservative delegation level, regardless of any system-side trust assessment.
- Let users set their own delegation preferences independently of system recommendations.
- Guarantee the non-agent channel genuine parity of service quality and timeliness, so withdrawing the agent never costs the user the service.
Response surface
The delegation widens as the agent earns trust and contracts the moment the user asks, with leaving no harder than joining.
Available after 12 confirmed actions. Nothing changes unless you accept.
Withdrawal is penalty-free by design: the non-agent channel runs at equal service quality, so the person fleeing a misbehaving agent is not the person the exit punishes.
Maturity
For structured earned-trust frameworks, proposed for enterprise agent contexts by vendor and preprint sources with no adopted standard or documented deployment cited here, and not yet adapted to government-user interactions, where the power asymmetry differs from employer-employee or business-customer relationships.
Precedents
CSA Agentic Trust Framework maturity model. The Intern, Junior, Senior, and Principal rungs treat trust as progressive and evidence-based: an agent earns greater autonomy through demonstrated trustworthiness rather than time served, and can be demoted when it violates a trust boundary. Demotion is written into the model at every rung.
CDR 90-day notifications (Australia). Under CDR Rule 4.20, once 90 days have passed since the consumer gave a consent, amended it, used their dashboard, or was last notified, the data recipient must notify them that the consent is still current, naming who holds it, for what purpose, and how to review it, and must deliver that notice outside the consumer dashboard. Standard consents are capped at 12 months. A standing authority the person has stopped thinking about is treated as a state to be surfaced.
Identity management for agentic AI. The paper binds a trust or delegation state to agent identity, so a level that expands and contracts stays attached to the agent it describes. It is a proposal, with no deployment behind it.
What carries over to agent use
High transferability. Pairing trust that an agent earns through use with delegation a user can revoke instantly maps directly to government AI agents, and both halves have precedent (earned-trust maturity models in enterprise governance, instant revocability in open banking). A user who has successfully used an agent to check payment dates might let it pre-fill a renewal form. If the pre-fill contains errors, they should be able to narrow the agent's scope at once or exit to a non-agent channel entirely.
Key design requirement: revoking delegation must never result in service degradation. A user who chooses not to use the AI agent must receive the same service quality and timeliness as one who does. Otherwise, graduated delegation becomes graduated coercion.
Where things go wrong
Where this goes wrong is a person caught in an erroneous automated process with no exit. Instant, penalty-free revocability and conservative defaults give that person a way to withdraw the agent and reach a human without losing service. The guarantee itself can be undermined without being formally broken: a system can re-offer the expanded scope on the next visit as though the revocation never happened, or bury the revoke control deep enough that only a persistent user finds it.
Sources
3 references
The instrument, the operating deployment, or the official record itself.
Writing about the subject rather than the framework itself, including vendor commentary.