Slide 4 · The Definition, Part 2
The second half of OWASP's definition is about you, not the model.
Consumers should be aware of how to interact safely with LLMs. They need to understand the risks of unintentionally providing sensitive data, which may later be disclosed in the model's output.
This is the half most teams skip. LLM02 isn't only “the AI leaked something.” It's also “someone fed it something they shouldn't have, and now it's part of what the model can surface later.”
Inadequate data sanitization
Overly broad data access for the model
No output filtering before the response is shown
Pasting confidential code into a chatbot
Treating an AI assistant like a private notebook
Not knowing whether input is used for training
One Line
Both sides of this failure are covered in Part 4. Some mitigations are things builders configure. Others are things users need to be taught.