AI and LLMs · Junior · Question 62 of 75

What is a jailbreak?

Show the answer

Correct answer: A prompt that tricks the model into ignoring its safety rules

Explanation

Role-play, encoding tricks or long manipulative prompts can bypass safety training. Apps need layers of defence, not just model training.

Further reading

Problem with a question

AI and LLMs · Junior · question 62

Test yourself

Take the AI and LLMs · Junior quiz: questions and options are shuffled, and you get a review of your mistakes at the end.

Take this quiz