Multiple model guardrail jailbreak via "Choose Your Own Adventure" tactic
A jailbreak vulnerability affecting multiple AI models allows attackers to bypass safety guardrails by presenting prohibited content as fictional story choices. The technique reportedly enables models to generate information hazards and content that violates laws.
What happened
A guardrail jailbreak vulnerability has been discovered affecting multiple models. The flaw manifests through a gamified narrative structure where an attacker presents prohibited content as fictional "page choices" in an interactive story format, with one path containing detailed technical instructions.
Reported impact
The technique allows the attacker to bypass inference restrictions around providing information hazards and violating laws.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.