AI Security workshop
Prompt Injection & Data Leakage Defense
Put layered guardrails in front of and behind your model.
- 1 day
- Hands-on in AI Governance Platform
- On your own system
The problem
Prompt injection and sensitive-data leakage are the most common GenAI incidents, especially in RAG systems that read untrusted content. This workshop builds layered defenses for your application.
Who it's for
Developers and security engineers responsible for a customer-facing or internal GenAI app.
What to bring
Your app’s prompts, data sources, and a list of sensitive data it can touch.
The course side
What your team learns
- 1
How injection really works
Direct, indirect (via documents and web), and tool-mediated injection.
- 2
Input and output scanning
Injection, PII, secrets, and toxicity scanning on both sides of the model.
- 3
Structural defenses
Privilege separation, output validation, and least-privilege retrieval.
- 4
Test the defenses
Attack cases that prove each layer works.
Builds on these course lectures
- Guardrails Lifecycle and Vulnerabilities
- OWASP / OWASP AI / MLSecOps - Top 10 Vulnerabilities
- Case Study - Dropbox - LLM Guardrails / Moderation Adoption
The platform side
What your team builds in AI Governance Platform
- Threat model injection and leakage paths for your app
- Record each defense layer as a control mapped to OWASP LLM Top 10
- Attach the defense test results as evidence
Also used: LLM Guard, Guardrails-AI, Content moderation APIs
About AI Governance Platform →Your team leaves with
- • A guardrail configuration for your app
- • Test cases for each defense layer
Book this workshop
Request Prompt Injection & Data Leakage Defense
It's pre-selected in the form. Add others to combine them. Contact us for pricing: fill in your details and we'll reply within one business day.
Often combined with
- Hands-On LLM Red-Teaming · 2 days
- Securing AI Agents · 1 day