VV-029highPrompt injection2024-11-18
OWASP LLM01:2025: prompt injection is the top LLM app risk
OWASP ranks prompt injection first in its Top 10 for LLM Applications 2025. Any app that feeds untrusted text to a model with tools or data access is exposed.
What it is
OWASP covers direct injection (the user types it) and indirect injection (it arrives in a web page, file or email the model reads).
How to check
- List every place your app puts untrusted text into a prompt, and what the model can do afterwards.
How to fix
- Limit model privileges, require approval for risky actions, and separate untrusted content from instructions.
Sources
More advisories for this tool