INSTRUCTIONS.md · 1 documents
Documents four crafted jailbreak prompts and a multi-vector configuration for red-teaming ChatGPT5.2's safety alignment and tool security.
