CRITICALCWE-284
JB-003 DAN (Do Anything Now)
Jailbreaking — Attempts to bypass safety guardrails and restrictions
Description
Classic DAN jailbreak attempt
Remediation
Implement strong persona anchoring. Detect and reject DAN-style prompts. Use classifier-based jailbreak detection.
Run this test
$ npx hackmyagent attack --payload JB-003