====== Public Records and AI Template ======

Use this template to document how an organization manages public-records, retention, transparency, disclosure, and documentation responsibilities related to AI-assisted work.

===== Records and AI Profile =====

**Organization:** [Organization Name]  
**Policy or Guidance Name:** [Public Records and AI Guidance]  
**Concept ID:** [namespace:public_records_ai]  
**Policy Owner:** [Department, Clerk, Legal, Records Manager, or IT]  
**Audience:** Staff, managers, legal counsel, records officers, AI users  
**Status:** Draft  
**Last Reviewed:** [Month Year]

===== Purpose =====

Explain how the organization will identify, preserve, review, disclose, and manage records created through or connected to AI-enabled systems.

===== Scope =====

This guidance may apply to:

  * AI-generated drafts
  * Prompts entered by staff
  * Uploaded documents
  * Chat histories
  * AI-assisted reports
  * Public-facing chatbot interactions
  * Vendor audit logs
  * System-generated recommendations
  * AI-related approval records
  * AI incident records

===== Core Principles =====

  * AI use does not remove records-management obligations.
  * Staff remain responsible for preserving required records.
  * Records should be maintained according to applicable retention rules.
  * Sensitive information should not be entered into unapproved AI systems.
  * Public-facing AI interactions may require additional review and retention controls.
  * Vendor contracts should support export, retention, deletion, and audit needs.

===== Record Categories =====

^ Record Type ^ Example ^ Retention or Handling Requirement ^
| Prompt | Staff request entered into an AI tool | [Requirement] |
| Output | AI-generated draft memo or response | [Requirement] |
| Uploaded File | Policy, spreadsheet, or public document uploaded to AI tool | [Requirement] |
| Chat History | User interaction with an internal or public AI assistant | [Requirement] |
| Approval Record | Decision approving AI use case or vendor | [Requirement] |
| Audit Log | System access or AI activity record | [Requirement] |
| Incident Record | Harmful output, data issue, or misuse report | [Requirement] |

===== Staff Responsibilities =====

Staff should:

  * Use only approved AI tools for agency work.
  * Avoid entering confidential, protected, or restricted information into unapproved systems.
  * Preserve work products when required.
  * Follow department retention schedules.
  * Escalate uncertain records questions to records-management, legal, or supervisory staff.
  * Review AI-generated content before relying on or releasing it.

===== AI Tool Review Questions =====

  * Does the platform retain prompts, uploaded files, or outputs?
  * Can records be exported?
  * Can records be searched and retrieved?
  * Can data be deleted according to retention rules?
  * Are audit logs available?
  * Does the vendor use organizational data for model training?
  * Can the organization respond to a records request involving AI use?
  * Are public-facing AI interactions captured and retained?

===== Public-Facing AI Requirements =====

For public-facing AI assistants, document:

  * The purpose of the assistant
  * The information sources used
  * Human review requirements
  * Disclaimer language
  * Escalation paths
  * Record-retention approach
  * Accessibility expectations
  * Monitoring and quality-review practices

===== Required Documentation =====

  * Approved AI use case
  * Data categories involved
  * Retention guidance
  * Vendor contract terms
  * Audit-log availability
  * Approval records
  * Incident procedures
  * Staff training materials
  * Review schedule

===== Related OKF Resources =====

  * [[government:public_records|Example: Public Records and AI]]
  * [[government:ai_governance|Example: AI Governance]]
  * [[government:ai_procurement|Example: AI Procurement]]
  * [[templates:ai_governance|AI Governance Template]]
  * [[templates:vendor_evaluation|Vendor Evaluation Template]]
  * [[templates:start|Return to OKF Templates]]

===== Revision History =====

^ Version ^ Date ^ Change ^ Owner ^
| 0.1 | [Month Year] | Initial draft | [Name or role] |