Workshop proposal · In preparation
Introduction to AI Application Security
90 minutes · Beginner IT learners · Controlled lab demonstration only after technical verification
Objectives
Recognize key parts of an AI application; distinguish authentication from authorization; identify confidential information and untrusted content; explain why model output is not an authoritative access-control decision.
Proposed 90-minute agenda
- 15 minutes — AI assistant architecture and data movement
- 20 minutes — Actors, assets, trust boundaries and authorization
- 20 minutes — Confidential prompts, retrieved documents and secrets
- 20 minutes — Untrusted inputs and a safe local demonstration
- 15 minutes — Questions, recap and short assessment
Prerequisites and practical activity
General familiarity with web applications is helpful but not mandatory. Practical activity uses a sandboxed, synthetic-data example to describe an input, enforcement boundary and observed result. No third-party systems or live targets will be tested.
Readiness and limitations
This syllabus is a proposal, not a course that has already been delivered. Lab setup, test results, slide content, learner accessibility and teaching rehearsal must be checked before booking. No unverified security finding will be described as demonstrated.
Proposed by: Ahmed Amhdour, Agadir · Teaching profile · Workshop inquiry
Résumé en français : Atelier proposé d'initiation à la sécurité des applications IA, avec un exercice pratique contrôlé après validation technique.