Trust & policies
AI Disclaimer
How to interpret AI outputs, Operators, Mission Replay, Mission Physics, Quality Gates, and user responsibility.
Final legal review required before public launch. This page is professional draft policy content and must be reviewed by qualified counsel before public launch.
AI outputs
AI Providers can produce incorrect or incomplete content
AI-generated content may be inaccurate, incomplete, outdated, biased, unsafe, unsupported, or unsuitable for your specific use case.
opheli.ai structures AI work into Objectives, Context, Operators, Runs, Artifacts, and Replay, but it does not guarantee correctness, completeness, safety, legality, or business outcomes.
Operators are AI/system roles or configured product roles. They are not human experts unless explicitly identified as human participants.
Review required
Every Artifact should be reviewed before external use
Users must review final Artifacts, cited evidence, assumptions, recommendations, and risk notes before relying on or publishing outputs.
Outputs may contain hallucinations, unsupported claims, incorrect citations, omitted caveats, or overconfident language.
Traceability
Mission Replay and Mission Physics are support mechanisms, not certification
Mission Replay shows execution trace and proof-of-work style signals. It is not a guarantee that every claim is true.
Mission Physics, Mission Stability, Quality Gates, Risk Contracts, and review indicators are scoring and review-support mechanisms. They are not certifications, legal assurances, compliance findings, or guarantees.
High-stakes use
Do not use output as the sole basis for high-impact decisions
Do not use opheli.ai output as the sole basis for legal, medical, financial, tax, employment, insurance, credit, housing, safety, compliance, or other high-impact decisions without qualified human review.
Provider dependency
Output quality depends on Provider, model, prompt, Context, and setup
The quality and behavior of outputs depends on the selected Provider, model, Provider availability, Operator setup, attached Context, prompt quality, and user instructions.