Automation governance
AI and Automation Policy
Draft rules for AI-assisted research, writing, images, checks, routing, data refreshes, and autonomous workflows.
Effective date: 2026-08-29
Review status: Owner execution version. Contact CREN for notices; not attorney approval.
Allowed automation
- Topic discovery, public-record monitoring, source collection, draft structuring, link checks, image-generation assistance, duplicate checks, data refreshes, queue triage, and internal reporting.
- Deterministic publication checks, image integrity checks, market-data quality status, broken-link audits, and policy scanners.
Human approval points
- Do not autonomously publish high-risk accusations, legal pages, corrections decisions, sponsored claims, rankings, investment language, fair-housing-sensitive recommendations, or final ad approvals.
- Material legal, financial, tax, lending, insurance, real estate, securities, safety, or reputation-sensitive claims need qualified human review.
- External outreach, purchases, credential changes, production deploys, payments, and public launches require explicit approval.
Disclosures and records
- AI-generated editorial images should be labeled when they represent a scene and are not documentary photos of a specific place or property.
- AI-assisted articles should preserve source links, source status, fact-check timestamps, image records, and correction context.
- Automation should create audit logs that identify actor, action, source, changed fields, and timestamp.