# splitting

Lab: conversation splitting-aanval

Split adversarial payloads across multiple conversation turns to bypass per-message safety classification.