Iterative Self-Tuning LLMs for Enhanced Jailbreaking Capabilities