Technology term
Multi-Turn Jailbreak
A multi-turn jailbreak is an adversarial conversation that attempts to bypass an AI system's safety rules gradually across several exchanges. Instead of presenting one explicit harmful request, the attacker divides the objective into smaller steps, adapts to refusals, and uses prior context to make later requests appear acceptable.
English
Multi-Turn Jailbreak
Arabic
كسر حماية متعدد الجولات
First appeared in
AI Agents Were Easier to Manipulate One Harmless Step at a Time
