Analysis

What Is LLM Jailbreak? Definition & Examples

Page published

Publication date unknown · First observed: 15 Nov 2026

Coverage timeline

15 Nov 2026nhimg.orgobserved

Single-source analysis — one report is available.

Why it matters

LLM jailbreaks can undermine refusal-based safeguards in deployed agents, making independent tool permissions, context isolation, and output monitoring important defensive controls.

NHI Mgmt Group’s glossary entry explains LLM jailbreaks as prompt-level attempts to bypass model safety policies through role-play, conflicting instructions, and context manipulation. The entry discusses potential data disclosure and connected-tool misuse, but presents no new exploit demonstration or observed incident.