Obedience
DocumentationDownload PDF
Responsible AIPublic guide

Understand responsible AI use

Know when AI assistance may be used, what human review remains mandatory and what information should not be submitted.

Task
Use the understand responsible ai use process consistently and retain the right evidence.
Audience
Customers, evaluators and governance teams
Owner
Obedience Governance
Version
2026.08
Reviewed
Next review

Purpose and scope

Know when AI assistance may be used, what human review remains mandatory and what information should not be submitted.

Use this guide as orientation, then rely on the signed agreement, current policy or named support response where those sources set a more specific obligation.

Prepare the right information

  • The AI-assisted task and intended decision.
  • The permitted data classification.
  • A named human reviewer.

Follow the controlled process

  1. Frame the needState the desired outcome, affected product and accountable contact. For understand responsible ai use, avoid assumptions that have not been confirmed in writing.
  2. Review and decideUse the current governed source, record material questions and obtain approval from the person who owns the decision.
  3. Retain evidenceKeep the final reference, date, decision and any follow-up action together so a later reviewer can reconstruct what happened.

Expected record

  • A bounded assistance use case.
  • Human review and provenance evidence.
  • A clear choice to apply, revise or reject output.

Security and disclosure boundary

  • Never send passwords, authentication codes, private keys or recovery codes to Obedience staff.
  • Share the minimum customer, project and commercial information needed for the stated purpose.
  • Public documentation explains controls and responsibilities without publishing exploitable topology, credentials or confidential implementation detail.