Contain: bring AI under company rules

1. 3.1Where AI skills live
   • AI skills live on platforms such as Claude, ChatGPT and Copilot.
   • Gemini has Gems, which do not connect well to apps (as of Nov 2026).
2. 3.2Choose one platform for the team: Claude or ChatGPT
   • One AI subscription is enough to start, so the choice matters: it is where the affordability comes from.
   • Claude and ChatGPT are the world-leading platforms: safe, and accessible to any company.
   • The curriculum's recommendation: put your team on Claude.
3. 3.3Why company work stays off personal accounts
   • Company accounts keep company data under company control, and keep it from being used to train AI models.
   • Security settings to turn on, and data training settings to check.
4. 3.4The four pillars of safe AI
   • Data: the platform's privacy policy and how it treats your data. It matters, and it is the smallest of the four.
   • Access: what you let AI reach, through connectors and on your computer.
   • Correctness: what sources AI draws on, and whether it is making anything up.
   • Accountability: who checks the output and owns it before it goes anywhere.
   • Three of the four are led by people, not the platform. Choosing a platform does not by itself make AI safe.
5. 3.5Rules for using AI at work
   • A company AI policy built on the four pillars.
6. 3.6Myths to clear up
   • Such as "Copilot is our only option" and "choosing one platform makes us safe."

A team account on one platform, and rules for using AI at work.