AI and LLMs · Middle · Question 19 of 75
What is insecure output handling in LLM apps?
- 1Using model output unchecked, e.g. rendering it as HTML
- 2Saving model output to logs without timestamps
- 3Showing model output in a font that's too small
Show the answer
Correct answer: Using model output unchecked, e.g. rendering it as HTML
Explanation
Rendering output with innerHTML can cause XSS; passing it to a shell or SQL can cause injection. Treat it like user input.