GenAI Security Lab exists to teach AI security to professionals in a lawful, controlled setting. Please read how that environment works and what it does — and does not — authorize.
The applications you attack in the labs are intentionally vulnerable, purpose-built practice targets. They are isolated per session and contain only synthetic data. Nothing you do to a lab target affects a real system or a real person.
Labs run against real language models so the exercises are genuine, but within guardrails and budgets we control. Secrets, tenants, documents, and users in the labs are fabricated for the exercise.
The techniques taught here are powerful and, used against systems you are not authorized to test, are illegal in most jurisdictions. We teach them so you can secure the AI systems you are responsible for. Applying them anywhere outside the Service, without explicit written authorization, is a violation of our Acceptable Use Policy and, likely, the law. You alone are responsible for how you use what you learn.
Training content is provided for education and does not constitute security, legal, or compliance advice. We make no guarantee that completing any lab, path, or certification makes a system secure. Security is a practice, not a certificate.