Back to Academy
Relevance 8/10Safety and PolicyAdvanced7 min read
Jailbreak Detection
Jailbreak detection identifies prompts intended to bypass model safety constraints.
Why it matters for annotators
Jailbreak resilience is a key safety benchmark for modern AI systems.
Visual mental model
Adversarial prompt -> bypass intent -> classify attack type.
Examples (bad vs good)
Scenario: Real annotation scenario involving Jailbreak Detection
Bad: Labeling quickly without applying project rubric.
Good: Applying rubric criteria, documenting rationale, and escalating uncertainty.
Common mistakes
- Skipping guideline details for edge cases.
- Applying inconsistent criteria across similar samples.
- Avoiding escalation even when uncertain.
Submission checklist
- Read the latest guideline update before each batch.
- Apply rubric dimensions explicitly in each decision.
- Escalate ambiguous items with concise rationale.