events
Make an agent safer to operate: permissions, tests and traces
Workshop online di 90 minuti su come rendere più sicuri gli agenti di IA che usano strumenti, con una matrice delle autorizzazioni, test negativi e una revisione delle tracce in una sandbox sintetica.

In inglese
- Starts
- 2026-12-13 17:00 UTC
- Organizer
- AI for Everyone Toronto | Beginner to Expert · AumOS
- Location
- Online event
A 90-minute live online session from AumOS, a community-led initiative, on operating tool-using AI agents more safely. Participants work through how such a workflow can fail, then set explicit boundaries: a permission matrix, negative tests and operator evidence. The format combines a worked example, an exercise, discussion and questions, with all tool actions confined to a synthetic sandbox. It is pitched at engineering, security and operations practitioners, assumes only basic familiarity with APIs or software workflows, and is free to attend.
Event website