Privacy and security risks · Question 7 of 60
What is the goal of context manipulation with extremely long prompts?
1 answer is correct. Easy Created with AI
Show answer
Answer
- C The model is meant to reveal fragments of its training data
Explanation
If an attacker exceeds the context window with very long inputs, the model may output random excerpts from its training data. These can include confidential details.