From an honest stocktake of AI tools in use to practical, jointly adopted team rules. Learn more: AI governance workshops.
Make clear from the start: this is about practical team rules, not a blanket ban on AI use. That lowers the barrier to honest contributions during the stocktake.
Collect openly and without judgment which AI tools are already in use across the team – including informal or unapproved ones. Honest answers only happen without the threat of sanctions.
Consider an anonymous collection method (see our post on decoupled brainstorming)
Build an approved list instead of a ban list: which tools are cleared for which purposes, and how does a new tool get added? A pure ban list, in practice, mostly just drives more shadow usage.
Define which data – customer data, source code, trade secrets – may go into which kind of tool at all. This often reveals that the real question is an unresolved cloud and data-protection issue that predates AI tools entirely.
Clarify when AI-generated content needs to be disclosed and who is responsible for reviewing its quality.
Define who stays substantively responsible for AI-assisted outputs. Responsibility never shifts onto the tool itself.
Treat the rules you've drafted as a concrete proposal and adopt them via an objection round, instead of waiting for full consensus across the whole organization (see our post on facilitating consent decisions).
The rules are explicitly provisional. Fix a date roughly three months out to adjust them based on actual usage and new tools.