Responsible AI
Responsible AI Policy
Grounded answers, scoped context, and human accountability — the rules behind every AI surface we ship.
Last updated: July 2026
Grounded, not invented
Assistants answer from verified context, not open-ended model knowledge, and say so when an answer isn't available rather than guessing.
Scoped by architecture
Access is enforced server-side — a surface only ever receives the context it needs, not whatever a prompt happens to allow.
Human accountability
AI-generated research is reviewed before publication. Product owners remain responsible for what their products publish.