Educational companion dossier · Fact, interpretation, lived experience, clinical education, fiction, and mechanics are labeled separately. Scope & safety
PsychologicalWar.orgPeople before labels.
REAL-WORLD INTERPRETIVE

AI-DRIVEN WORLD SAFETY

AI NPC Identity, Consent, Memory, and Provenance

A clear boundary for AI-driven characters: players know when generative systems are involved, what data is used, what the NPC remembers, and which outputs can change game state.

Read the working brief Choose a learning path
LEVEL 1

ORIENTATION

Why this matters

REAL-WORLD INTERPRETIVE

One-sentence brief

Human-like dialogue can enrich a world, but hiding controller type, impersonating real people, or using intimate data without consent undermines agency and trust.

REAL-WORLD INTERPRETIVE

Three key points

  1. AI-mediated characters need persistent, accessible identity signals.
  2. Memory must be inspectable, correctable, scoped, and deletable where policy allows.
  3. Generative output cannot directly control sanctions, assets, diagnosis, or canonical truth.
LEVEL 2

WORKING BRIEF

Evidence, context, and limits

REAL-WORLD INTERPRETIVE

Disclose AI involvement

Use an always-available AI or synthetic-character indicator, with optional diegetic presentation only when the player has already consented to that fiction. Do not make players prove whether an interlocutor is human.

  • Text and screen-reader labels are required.
  • Voice or visual style alone is insufficient.
  • Never impersonate a real player or staff member.
REAL-WORLD INTERPRETIVE

Memory controls

Separate canonical character facts, session context, player-provided preferences, and sensitive disclosures. Give players understandable controls to review, correct, limit, or delete personal memory where compatible with safety and legal obligations.

  • Do not store more than needed.
  • Never convert health disclosure into a suspicion score.
  • Record source and last update for character facts.
REAL-WORLD INTERPRETIVE

Authority boundary

The model may propose dialogue or actions. The authoritative engine checks permissions, location, ownership, objective state, safety policy, and economic consequences. Rejected proposals should fall back safely without repeated costly retries.

  • No direct currency minting.
  • No autonomous bans or diagnosis.
  • No hidden transfer of private data between rooms.
REAL-WORLD INTERPRETIVE

Provenance and correction

Generated claims about lore, missions, other players, or institutions should carry source state and confidence. A correction should update future dialogue and any derivative quest or record.

  • Preserve the original and corrected state.
  • Avoid fabricated citations.
  • Show when the character is uncertain or does not know.
LEVEL 3

COMPLETE DOSSIER

Limitations, game links, and review context

DISPUTED / MULTIPLE ACCOUNTS

Known limitations and gaps

  • The supplied reports are preserved research inputs, not independent proof of every cited case, statistic, legal claim, or product description.
  • Public material abstracts system design and player-protection principles; it omits actionable intrusion, evasion, coercion, targeting, sabotage, or real-person profiling methods.
  • Player behavior, diagnosis, disability, nationality, religion, language, poverty, migration, or social identity are never automatic indicators of guilt, fraud, danger, or disloyalty.
  • Parameters require simulation, accessibility testing, privacy review, regional review, and human playtesting before production use.
LEVEL 4

RESEARCH EDITION

Sources, methods, and stable links

REAL-WORLD INTERPRETIVE

Linked reports

  • AI-Driven NPCs in Adult GamesProtected research reference; no public report route in this release.
  • AI NPC Cost Architecture PlanProtected research reference; no public report route in this release.
REAL-WORLD VERIFIED

Method and corrections

This page follows the public method for provenance, confidence, source independence, alternative accounts, limitations, review state, and visible correction.

Read the methodView corrections
NEXT

CONTINUE

Related learning

AI-Driven World SafetyAI Mission Generation: Reliability, Verification, and Fail-ForwardA mission pipeline in which generative systems propose bounded narrative variations while deterministic validators ensure that objectives, locations, rewards, permissions, and exits are real.AI-Driven World SafetyAI NPC Cost, Latency, and Graceful DegradationA bounded architecture for generative characters that protects budget, responsiveness, privacy, and authoritative game state under normal play, spikes, outages, and abuse. Learning pathAI-Driven NPC Safety and ReliabilityKeep AI identity visible, game state deterministic, missions valid, memory scoped, fallbacks safe, and player wellbeing central.