# OWASP Top 10 for LLM Applications

> A free list of the ten biggest risks in apps built on LLMs.

**Field:** Artificial Intelligence  
**Category:** Safety & Security  
**Also known as:** OWASP LLM Top 10, OWASP Top 10 for LLMs, OWASP GenAI Top 10

## Think of it like

It is like a list of the ten most common ways burglars get into a house. You check your house against it.

## Definition

The OWASP Top 10 for LLM Applications is a free list of the ten biggest risks in apps built on large language models. It comes from OWASP, a group known for its lists of web risks. The 2025 list starts with prompt injection. Next comes leaking private data. It also covers poisoned data, unsafe use of a model's output, agents with too much power, leaked system prompts and runaway costs. Teams use it as a checklist when they build and test an AI app.

## Example

Emma's team checks their new chatbot against the 2025 list. Item LLM06 is about too much power. So they take away the bot's power to give refunds by itself.

## Why it matters

It gives builders a shared list of what to test, written by people who see real attacks.

## Common mistake

Thinking the web Top 10 covers AI apps. The LLM list adds risks a normal web app does not have.

## Three principles

- **Ten top risks.** Ranked for apps built on LLMs.
- **Free and open.** Made by a nonprofit.
- **A checklist.** Test your app against it.

## Related terms

- [OWASP Top 10](https://learn.benibix.com/term/development/owasp-top-10)
- [Prompt Injection](https://learn.benibix.com/term/ai/prompt-injection)
- [Excessive Agency](https://learn.benibix.com/term/ai/excessive-agency)
- [Prompt Leaking](https://learn.benibix.com/term/ai/prompt-leaking)
- [Data Poisoning](https://learn.benibix.com/term/ai/data-poisoning)
- [AI Security](https://learn.benibix.com/term/ai/ai-security)

![The ten items of the 2025 OWASP Top 10 for LLM Applications in two columns: prompt injection, leaking private data, supply chain, data and model poisoning, unsafe use of output, excessive agency, system prompt leakage, vector and embedding flaws, misinformation, and runaway use and cost. Prompt injection, excessive agency and system prompt leakage are highlighted.](https://learn.benibix.com/diagrams/ai4-owasp-llm.svg)

---

Source: https://learn.benibix.com/term/ai/owasp-top-10-for-llm-applications
Author: Behnia Zarei (a developer in Vienna, Austria, who builds and teaches data, AI and business intelligence at BENIBIX)
Publisher: BENIBIX Learn (https://learn.benibix.com/)
Last updated: 2026-11-18
Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Quote freely with attribution.
