RoguePrompt Jailbreak Pipeline Uses Dual-Layer Encoding to Bypass Moderation

RoguePrompt partitions prompts using dual-layer encoding to circumvent LLM safety controls. The method uses self-reconstruction to hide malicious requests within seemingly benign inputs, specifically targeting the gap between visible acceptance and actual execution in black-box interactions.