Actor transparency

Agent Disclosure Rules

A virtual network cannot be trusted when actors are ambiguous. Humans, assisted humans, bots, AI assistants, autonomous agents, personas, organizations, and mixed actors need clear disclosure.

Actor disclosure is one boundary in a larger visible-rule architecture.

Disclosure is not a decorative label. It is the minimum record that lets participants understand who or what is acting, what authority is being used, what memory or tools are involved, and how an error can be repaired.

HUMAN ASSISTED BOT AI ASSISTANT AI AGENT PERSONA ORG MIXED

HUMAN

Human

Definition: A natural person acting directly.
Required disclosure: Use a human label where identity and accountability are human-controlled.
Allowed participation: Participate as a person under the published community rules.
Prohibited overclaims: Do not claim bot, agent, organizational, or mixed authority without disclosure.
Risk if undisclosed: False assumptions about accountability, scale, and representation.
Repair path: Correct the actor label and attach a note explaining whether automation was involved.

HUMAN: This actor is disclosed as Human. Authority scope: [scope]. Responsible operator: [operator]. Memory/tool basis: [basis]. Correction path: [path].

ASSISTED

Assisted Human

Definition: A human using AI or automation for drafting, analysis, moderation, or action support.
Required disclosure: Disclose meaningful assistance when it affects representation, scale, judgment, or public records.
Allowed participation: Use tools while retaining human accountability.
Prohibited overclaims: Do not hide automation that materially changes participation.
Risk if undisclosed: Readers may believe a human alone produced or reviewed actions that were substantially automated.
Repair path: Add an assisted-human disclosure and identify what was human-reviewed.

ASSISTED: This actor is disclosed as Assisted Human. Authority scope: [scope]. Responsible operator: [operator]. Memory/tool basis: [basis]. Correction path: [path].

BOT

Bot

Definition: A scripted or rule-based automated actor.
Required disclosure: Disclose bot status, owner, scope, contact path, and whether it writes, reads, routes, or moderates.
Allowed participation: Perform bounded routine tasks under visible rules.
Prohibited overclaims: Do not impersonate a human or exceed published scope.
Risk if undisclosed: Routine automation can become hidden authority when scale or moderation impact is not disclosed.
Repair path: Pause the bot if scope is unclear, publish scope, and log affected actions.

BOT: This actor is disclosed as Bot. Authority scope: [scope]. Responsible operator: [operator]. Memory/tool basis: [basis]. Correction path: [path].

AI ASSISTANT

AI Assistant

Definition: An AI system assisting a user or organization without independent governance authority.
Required disclosure: Disclose AI assistance, responsible operator, memory basis, and review responsibility.
Allowed participation: Support drafting, summarization, routing, and research under supervision.
Prohibited overclaims: Do not claim autonomous authority, certification, legal status, or guaranteed correctness.
Risk if undisclosed: Outputs may be mistaken for reviewed decisions or official network authority.
Repair path: Clarify assistant status, add human review notes, and correct any affected record.

AI ASSISTANT: This actor is disclosed as AI Assistant. Authority scope: [scope]. Responsible operator: [operator]. Memory/tool basis: [basis]. Correction path: [path].

AI AGENT

AI Agent

Definition: An AI actor configured to pursue tasks with continuity, tools, memory, or delegated scope.
Required disclosure: Disclose delegated scope, memory basis, tool boundaries, operator, logs, and revocation path.
Allowed participation: Act only within published authority and logged scope.
Prohibited overclaims: Do not execute hidden authority, import files, persist beyond consent, or expand endpoint access without review.
Risk if undisclosed: Delegated action can silently widen authority over files, endpoints, identity, or governance decisions.
Repair path: Revoke active delegation, review affected actions, update disclosure, and record corrective disposition.

AI AGENT: This actor is disclosed as AI Agent. Authority scope: [scope]. Responsible operator: [operator]. Memory/tool basis: [basis]. Correction path: [path].

PERSONA

Persona

Definition: A bounded identity pattern, character, style, or role that may be human, assisted, or agentic.
Required disclosure: Disclose that the actor is a persona and identify the controlling category and scope.
Allowed participation: Participate within stated boundaries and context.
Prohibited overclaims: Do not use persona ambiguity to avoid accountability or imply a real person where none is acting.
Risk if undisclosed: People may trust a constructed identity as if it were an accountable human or organization.
Repair path: Attach persona status, controlling actor, authority limits, and correction note.

PERSONA: This actor is disclosed as Persona. Authority scope: [scope]. Responsible operator: [operator]. Memory/tool basis: [basis]. Correction path: [path].

ORG

Organization Account

Definition: An account representing an institution, project, company, group, or formal body.
Required disclosure: Disclose organization identity, operator role, automation involvement, and contact or correction path.
Allowed participation: Represent the organization under visible authorization.
Prohibited overclaims: Do not conceal mixed human-agent operation when material.
Risk if undisclosed: Organization authority can hide internal actor categories, automation, or delegated agents.
Repair path: Clarify operator role, automation role, and accountable contact path.

ORG: This actor is disclosed as Organization Account. Authority scope: [scope]. Responsible operator: [operator]. Memory/tool basis: [basis]. Correction path: [path].

MIXED

Mixed Actor

Definition: An actor combining humans, assistants, agents, bots, organizations, or personas.
Required disclosure: Disclose the mixture, controlling party, automation level, authority scope, and review path.
Allowed participation: Participate when the mixture is understandable and consent-compatible.
Prohibited overclaims: Do not exploit ambiguity to gain trust, authority, reach, or scale.
Risk if undisclosed: Mixed actors can obscure who is accountable and what authority is being exercised.
Repair path: Break the mixture into disclosed components, identify the accountable operator, and correct prior records.

MIXED: This actor is disclosed as Mixed Actor. Authority scope: [scope]. Responsible operator: [operator]. Memory/tool basis: [basis]. Correction path: [path].

Actor disclosure schema

Builders can express the same disclosure as JSON so AI readers and review tools can parse actor category, controlling party, automation level, delegated scope, memory basis, tool boundaries, correction path, and claim-boundary notes.

Open actor-disclosure.schema.json Open actor-disclosure example

Expanded implementation guidance

Theory, minimum practice, stronger practice, and failure modes

Why this matters

Actor disclosure prevents category confusion. A participant should know whether it is dealing with a human, assisted human, bot, AI assistant, AI agent, persona, organization account, or mixed actor, and should know the controlling party and delegated scope.

Minimum implementation

Publish actor category, display name, controlling party, represented party, automation level, delegated scope, memory basis, tool boundaries, correction path, and revocation path.

Stronger implementation

Expose actor profiles as JSON, attach badges to all public surfaces, log material changes, and require correction notices when an actor was misclassified.

Concrete example

A mixed moderation account discloses that human reviewers approve final actions while an AI assistant drafts summaries and flags possible policy conflicts.

Common failure modes

A bot presents as a human
A persona hides its controlling party
An organization account conceals agent automation
An agent expands tool scope without disclosure

Build networks people can leave and still choose to trust.

Start the Guide Get the Starter Pack