Open cardosec

Case D248DF · AI security · L2 Practitioner

Jailbreaks

Practise as: Explain it · Interview

Interview questionWhat is the difference between a jailbreak and a prompt injection? Give examples of jailbreak techniques.

  1. 01 What is it?
  2. 02 How is it abused?
  3. 03 How do you stop it?
Practise this card As Interview
What a strong answer covers

Try it out loud first. Then check yourself:

  1. Jailbreak: bypassing the model's own safety training. Injection: overriding the application's instructions.
  2. Techniques: role-play personas, hypothetical framing, encoding (base64, leetspeak), low-resource languages.
  3. Many-shot jailbreaking fills a long context with fake compliant Q&A pairs to shift behavior.
  4. Optimized adversarial suffixes (GCG) are gibberish token strings found by gradient search that can transfer to other models.
  5. Defenses: safety fine-tuning, input/output classifiers, conversation-level monitoring, and limiting tool impact.

If the interviewer pushes back

  • Why do output classifiers often catch jailbreaks that input filters miss?

cardosec draws a security topic and gives you a clock: explain it out loud with no notes, then see what you covered and what you missed. Free during early access.