AI glossary Responsibility
What is Jailbreaking?
← All glossary termsAttempts to evade a model's safety restrictions.
Explained
Jailbreaking refers to techniques used to bypass safety guardrails in AI models—to elicit harmful, biased, or restricted content. Robust AI governance includes monitoring and hardening against such attacks.