Glossary: Application and Credential Attacks, and AI Risks

OWASP Top 10 for LLM Applications

Also written: LLM Top 10, Top 10 for LLM

OWASP's ranking of the biggest risks for apps built on LLMs. Prompt injection is first, sensitive information disclosure is second, and improper output handling is also on the list.

Where the book teaches it

Chapter 4: Attacks on AI and Its Weak Spots
Chapter 5: AI Privacy, Ethics, Sessions and Code

Related terms

OWASP (Open Worldwide Application Security Project)prompt injection data loss improper output handling