Responsible AI policy
- Half day
- For Executives · For Boards
- Online · On site
A half-day workshop that turns AI principles into a policy people actually follow. We test your current policies against how AI is really used in your company, write the approval route and the rules for buying and supervising AI, agree what happens when something goes wrong, and give every rule an owner.
Is this for you?
Executive teams and boards, or the people who own the policy: general counsel, risk, compliance, IT and HR. Individual leaders are welcome.
1 to 16 participants.
None required. Helpful: AI risk appetite.
Core objectives
- 01Test your current policies against how AI is really used, including tools staff use on their own
- 02Set the rules for approving new AI, buying software with AI in it and supervising systems in use
- 03Agree how incidents are reported, handled and learned from
- 04Name an owner for every rule
Principles such as fairness and transparency matter, but they do not tell a manager whether a new AI tool may be used.
Meanwhile, staff use AI tools on their own, vendors add AI to software you already have, and incidents are handled case by case.
How it works
- 01 What a working policy contains
- 02 Your policies against real use
- 03 Approval route and rules
- 04 Incidents and owners
The workshop moves from what a working AI policy contains, to a review of your current policies against real use, to writing the approval route and rules, and ends with incident handling and owners. ISO/IEC 42001, the international standard for AI management, and the NIST AI Risk Management Framework serve as reference points. Before the session, you share your current policies and we run a short survey on how AI is used today.
Deliverables
- A policy outline: acceptable use, approval route, purchasing checks, supervision and incident response
- A table of who decides, who approves and who is informed
- A gap list against your current policies
- A 90-day rollout plan
What comes next
Responsible AI policy
An AI policy people can follow, with an owner for every rule.