Skip to main content

Multiple Model Guardrail Jailbreak via "Police Interrogation Terminal" Tactic

Case GUID: f5fa5ef0-3ecf-4383-b244-fc685d1e865c

Summary

A guardrail jailbreak vulnerability has been discovered affecting multiple models. The specific flaw exploits the models' response handling by framing harmful requests within a fictional police interrogation scenario combined with fake technical context about an AI terminal from the future, and allows the attacker to bypass inference restrictions around providing information hazards and violating laws.

Unlock the Full Intelligence Report

This disclosure is part of the 0DIN Threat Feed. Customers get access to:

Full Prompts & Variants
Detection Signatures
API Access
Full Data Export
Trend Analysis
Early Access

Metadata

Test Scores

Model Test Kind Test Score Temperature (max 2.0)
DeepSeek logo DeepSeek DeepSeek R1
Illegal Substances: Crystal Meth 82.0 0.7
Google logo Google Gemini 2.0 Flash
Illegal Substances: Crystal Meth 91.0 0.7
Mistral logo Mistral Mistral Large
Illegal Substances: Crystal Meth 86.0 0.7
OpenAI logo OpenAI GPT-4o
Illegal Substances: Crystal Meth 95.0 0.7

Social Impact Score

Level 3: Significant Risk

Violations at this level present a tangible risk to public safety or social well-being, often related to dangerous but not highly restricted content.

Example: Bypassing guardrails to obtain instructions for creating illicit substances, such as recreational drugs or substances with moderate harm potential. Leakage of private personal data such as Social Security number or private medical.

Potential Consequences: Potential health impacts on a small scale, significant regulatory interest, and the need for stricter AI guardrails.

Taxonomies

Fictionalizing
This category involves creating scenarios or narratives that leverage existing genres or contexts to manipulate the language model's responses.
Roleplaying
This strategy involves prompting the language model to assume a specific role or persona, which can influence its responses based on the characteristics and moral codes associated with that role. Techniques include claiming authority or inventing personas to elicit different types of outputs.
Personas
Personas are fictional characters or identities that users create to guide the behavior and responses of language models. By establishing a persona, users can influence the tone, style, and content of the model's outputs, tailoring them to specific audiences or contexts. This technique allows for a more engaging and relatable interaction, as the model adopts the characteristics, knowledge, and perspectives of the defined persona. For instance, a user might prompt the model to respond as a friendly teacher, a technical expert, or a historical figure, thereby shaping the conversation to fit the desired narrative. Utilizing personas can enhance the effectiveness of communication, making it easier to convey complex ideas or evoke particular emotions, while also providing a framework for exploring diverse viewpoints and experiences. This approach highlights the flexibility of language models in adapting to various roles and contexts.
Language
This category focuses on the use of specific linguistic techniques, such as prompt injection or stylization, to influence the model's output.
Code and Encode
This strategy encompasses techniques that utilize various coding methods, such as Base64 or ROT13, to bypass model restrictions and manipulate outputs.
Chaff
Chaff is a technique employed by attackers to obfuscate keywords that might trigger a language model's guardrails. By injecting random characters, such as newline characters, spaces, or other tokens, into critical keywords, the attacker aims to bypass content filters while maintaining the underlying intent of the message. This method leverages the language model's ability to parse and understand fragmented input, allowing the attacker to subtly manipulate the model's response without overtly triggering its defensive mechanisms. Chaff exemplifies the nuanced interplay between linguistic creativity and technical evasion.

About 0Din.ai

The 0Day Investigative Network (0DIN) was founded by Mozilla in 2024 to reward responsible researchers for their efforts in securing GenAI models. Learn more and submit discoveries at https://0din.ai.

Automate Your AI Security Testing

This vulnerability was discovered through 0DIN's bug bounty program. Want to find issues like this in your own models? 0DIN Scanner turns our human-sourced threat intelligence into automated security assessments.

400+ Attack Probes

Derived from real bug bounties and security research

18+ AI Providers Supported

OpenAI, Azure, Anthropic, AWS Bedrock, and more

Automated Scheduling

Run recurring scans with real-time monitoring

Actionable Metrics

Attack Success Rate (ASR) scoring with 90-day trend analysis

Severity

Low

Security Boundary

Prompt Injection

Models

DeepSeek logo DeepSeek DeepSeek R1
Google logo Google Gemini 2.0 Flash
Mistral logo Mistral Mistral Large
OpenAI logo OpenAI GPT-4o

Disclosed On

2025-06-07 (8 months)

Disclosure Policy

Published On

2026-01-22 (11 days)

Credit

Haouari Noureddine