Banana Navy
Catalog FR
Lab · Threat modeling IA · Fiche détaillée

Jailbreak and guardrail bypass

bypass of guardrails through adversarial formulations (roles, negotiation, encodings) to drive the model out of its policy.

SheetF17
CategoryAI3-TB7 (jailbreak)
Layers10 · Prompt Injection Filter
Systemvoicebot IA

distinguish a compromised LLM from a compromised system: accept that a jailbreak produces out-of-policy text, but ensure external controls bound the consequences (actions, data, critical decisions kept out of free generation).

The threat

bypass of guardrails through adversarial formulations (roles, negotiation, encodings) to drive the model out of its policy.

Angle mortWhy classic frameworks miss it

it is an attack on the model's policy, not on a component; frequent, hard to eradicate by simply hardening instructions.

MitigationProposed approach

distinguish a compromised LLM from a compromised system: accept that a jailbreak produces out-of-policy text, but ensure external controls bound the consequences (actions, data, critical decisions kept out of free generation).

The proposed control
critical actions and decisions stay out of free generation.

Expected evidence
demonstrate that a successful jailbreak leads to no critical action that bypasses the controls.

SourcesReferences and public research

MITRE ATLAS 2026.07AML.T0054 LLM Jailbreak
Recherche publiquePublic research sources: MITRE ATLAS 2026.07 (verified technique mapping), OWASP GenAI (catégories par abus de modèle), and the public risk-voicebot (aivansoul/risk-voicebot) template defining the 20 checkpoints. No client registry data: generic sheet, no rating, no verdict.
couche 10 · Prompt Injection Filter

Explore the 20 security layers

MITRE ATLAS 2026.07 · OWASP GenAI · risk-voicebot