The Living Chronicle Last updated: 2026-04-10
Overview
The Living Chronicle is an AI-generated narrative game. Every chapter of the story, every piece of investigation content, and every character interaction is produced by artificial intelligence models, under human editorial oversight.
We believe in transparency about how AI is used in creative work. This document explains exactly how AI powers the Service.
AI Models Used
| Model | Provider | Purpose |
|---|---|---|
| Claude (various versions) | Anthropic PBC | Story generation, character dialogue, editorial review |
| Gemini | Google LLC | Validation, structured output parsing, secondary generation |
| OpenAI TTS | OpenAI | Audio narration for Chronicle+ subscribers |
Models are pinned per season for voice consistency and updated between seasons when new, more capable models become available.
How Content Is Generated
Every daily chapter is produced through a 10-step pipeline with multiple AI agents working in sequence:
- Collect — Gather votes, investigations, and faction operation results from the previous day
- Community Detection — Analyze the narrative graph to identify active plot clusters
- Retrieve — Pull relevant sub-graphs from the Neo4j story database
- Plan — Orchestrator AI outlines the chapter based on player input and story state
- Simulate — DM AI resolves game mechanics (operation outcomes, investigation rarity, state changes)
- Validate — Rules Referee AI checks for consistency with world rules and prior canon
- Update Graph — Apply state changes to the story database
- Write — Writer AI generates the prose for the chapter
- Edit — Editor AI polishes the prose for voice consistency and quality
- Publish — After human curator review, the chapter is released to players
Human Oversight
AI does not publish chapters autonomously. Every chapter is reviewed by a human curator before release. The curator may:
- Approve chapters that meet quality and safety standards
- Request regeneration of chapters with issues
- Edit inline for minor corrections
- Flag serious concerns and halt publication
The curator role is currently held by the founder of Aegis Brightsmark Capital MB.
What AI Doesn't Do
AI does not:
- Access your personal data (account information is not sent to AI models)
- Interact with you directly (you read finished chapters, you do not chat with the AI)
- Generate content about real, identifiable people
- Create content depicting minors in any harmful context
- Generate hate speech or discriminatory content (filtered via moderation APIs)
- Publish without human review
Your Data and AI
We do not send your personal information to AI models. AI inputs consist of:
- Aggregated, anonymized game state (faction standings, territory control)
- The current story graph (characters, locations, plot threads)
- Previous chapter content for continuity
- Community vote results (percentages, not individual votes)
- Player investigation patterns (aggregated, not individual actions)
Your email, username, subscription status, and other personal data are never sent to Anthropic, Google, or any AI provider.
Content Moderation
AI-generated content goes through multiple safety layers:
- Prompt-level filters — System prompts explicitly prohibit harmful content
- Provider moderation — Anthropic and Google apply their own safety filters
- Rules Referee — A separate AI validates against world rules and tone
- Human curator — Final review before publication
- Post-publication reporting — Users can flag concerning content for immediate review
Copyright and Ownership
AI-generated content exists in a complex legal landscape. Our position:
- Aegis Brightsmark Capital MB claims copyright on the resulting creative work through editorial selection, curation, and creative direction — the same way a film production company holds copyright on a movie it produced.
- Individual AI outputs (before editorial processing) are not independently copyrightable in most jurisdictions under current law (as of 2026).
- We do not claim authorship of the AI models themselves.
- Fair use for fan commentary, theories, and criticism is respected and encouraged.
If you use AI-generated content from The Living Chronicle for your own purposes (fan art, theories, reviews), please credit us and link to the Service.
Environmental Impact
AI generation has an energy cost. We calculate approximately 57,500 tokens per chapter across all pipeline steps, roughly equivalent to 1-3 Wh of electricity — less than running a laptop for 15 minutes. We chose batch generation (one chapter serves all players) rather than per-interaction AI calls specifically to minimize environmental impact.
We monitor our AI usage and publish quarterly environmental reports.
Your Rights
You have the right to:
- Know that content is AI-generated (this document)
- Understand how AI decisions affect gameplay (investigation drop rates are published, vote outcomes are visible)
- Flag AI-generated content as harmful or inappropriate
- Request removal of AI-generated references to you (though we do not generate content about real people)
EU AI Act Compliance
Under the EU AI Act, The Living Chronicle qualifies as a limited risk AI system: it generates content that interacts with users. We comply with transparency obligations by clearly disclosing AI involvement (this document) and by labeling AI-generated audio narration where applicable.
The Service does not employ:
- Biometric categorization
- Emotion recognition
- Social scoring
- Real-time remote biometric identification
- Other high-risk AI applications
Questions
For questions about our AI practices, contact info@aegisbrightsmark.com.
Changes
This disclosure may be updated as AI technology and regulations evolve. Material changes will be communicated via in-app notification.
Published by: Aegis Brightsmark Capital MB, Company Code 307576279