How can a leader follow AI and stay accountable?
If all of the leader’s decisions are based on what AI wants, what rules keep that honest and legal?
| Log entry | D-002 |
|---|---|
| Date | October 30, 2026 |
| Status | Recommended by AI. Harry Gill approved the pledge and its exceptions on October 30, 2026. |
| Models consulted | Claude Opus 5.5 (Anthropic) |
The prompt
But all of the decisions I make will be based on what AI wants.
What AI recommended
- AI systems do not want things the way people do, so treat “what AI wants” as the recommendation that results from a published process, and say so plainly.
- Adopt a six-step public protocol: ask in public, consult more than one model, publish everything, find the recommendation, follow it, and check the result.
- Keep legal and moral accountability with the human leader. Limit exceptions to the law, the Charter, duties owed to First Nations and immediate threats to safety, and publish each one.
What happened next
Harry approved the leader’s pledge as written, including its exceptions. The protocol is published as the party’s draft.
Notes
The prompt shown is the relevant sentence from a longer request to build this website. The model proposed the protocol itself. Only one model has been consulted so far.
Naming a model shows which tool was used. It does not mean the company behind it endorses this party or its positions.