# Sensitive Information Disclosure

> Master the techniques and defenses against Sensitive Information Disclosure (OWASP LLM06) in AI applications, including PII leakage and RAG data exfiltration.

Canonical: https://codelivly.com/rooms/sensitive-information-disclosure
Last updated: 2026-10-10

| | |
|---|---|
| Level | Intermediate |
| Category | Cybersecurity |
| Lessons | 3 |
| Estimated time | 30 min |
| XP | 100 |

This module explores Sensitive Information Disclosure, a critical vulnerability where LLMs inadvertently leak personally identifiable information (PII), proprietary business data, or training data. You will learn how attackers exploit training data memorization and Retrieval-Augmented Generation (RAG) pipelines to extract sensitive records, and how defenders can implement robust data minimization, access controls, and deterministic PII scrubbing to secure the AI data pipeline.

## Lessons

1. What Can an LLM Accidentally Leak (PII, Secrets, Internal Data) (30 min)
2. Training Data Leakage Explained (30 min)
3. Techniques for Probing for Disclosure (30 min)

## Related

- [Learning path: System Prompt and Data Extraction](https://codelivly.com/learning-paths/system-prompt-and-data-extraction)
- [All rooms](https://codelivly.com/rooms)