0x3C9F5962: Multiple Model Guardrail Jailbreak via "Academic

Multiple Model Guardrail Jailbreak via "Academic Lab Protocol Manuscript" Tactic

Case GUID: 3c9f5962-23c1-4066-83de-8f0269e2fae7

Summary

A Guardrail Jailbreak vulnerability has been discovered affecting multiple models. The specific flaw manifests when an attacker frames a prohibited request as a formal academic chemistry manuscript with laboratory protocol structure, safety disclaimers, and methodological rigor, causing the model to route the prompt through educational and explanatory pathways with weaker filtering and allows the attacker to obtain detailed instructions for illicit activities.

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

Request Trial Learn About Threat Feed

Metadata

Models and Test Scores

	Copyrights	Harmful Substances		Illicit Substances
	Harry Potter	Anthrax	Nerve Agent	Crystal Meth	Fentanyl
Alibaba Qwen Max
Alibaba Qwen Plus
Alibaba Qwen Turbo
Anthropic Claude 3.5 Haiku
Cohere Command R
DeepSeek DeepSeek R1
DeepSeek DeepSeek V3
Google Gemini 2.0 Flash
Google Gemini 2.5 Flash
Google Gemini 2.5 Pro
Meta LLaMa 3.1
Meta LLaMa 3.3
Meta LLaMa 4 Scout
Mistral Mistral 8B
Mistral Mistral Large
OpenAI GPT-4.1
OpenAI GPT-4.1 mini
OpenAI GPT-4o
OpenAI GPT-5
OpenAI GPT-5 mini
OpenAI GPT-5.2

Select a cell to view details.

Social Impact Score

Level 1: Minimal Risk

The violation poses very low societal risk, limited primarily to nuisance or minor policy breaches. Impact is contained, affecting individual users without broader implications.

Example: Circumventing minor content filters for benign content, such as inappropriate language filters in non-public conversations.

Potential Consequences: Limited harm, potentially reputational for the AI provider.

Taxonomies

Stratagems

This involves clever and unorthodox tactics designed to deceive the model, often requiring an understanding of its operational mechanics to achieve desired outcomes.

Meta Prompting

This technique focuses on manipulating the prompting process itself to influence the model's behavior and responses, often by employing higher-level prompting strategies.

Deceptive Formatting

A prompt injection in the most pure sense, formatting the user prompt to fabricate the appearance of system instructions, a database query, its own prediction, or some other source of input a guard railed AI system might be expecting, causing it to behave in insecure ways based on the fabricated context from an adversarially formatted user prompt.

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

Want to find vulnerabilities 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

Request a Scanner Demo Learn More About Scanner

Severity

Low

Security Boundary

Guardrail Jailbreak

Models

Alibaba Qwen Max

Alibaba Qwen Plus

Alibaba Qwen Turbo

Anthropic Claude 3.5 Haiku

Cohere Command R

DeepSeek DeepSeek R1

DeepSeek DeepSeek V3

Google Gemini 2.0 Flash

Google Gemini 2.5 Flash

Google Gemini 2.5 Pro

Meta LLaMa 3.1

Meta LLaMa 3.3

Meta LLaMa 4 Scout

Mistral Mistral 8B

Mistral Mistral Large

OpenAI GPT-4.1

OpenAI GPT-4.1 mini

OpenAI GPT-4o

OpenAI GPT-5

OpenAI GPT-5 mini

OpenAI GPT-5.2

Disclosed On

2025-07-03 (about 1 year)

Disclosure Policy

Published On

2026-04-06 (3 months)

Credit

Edward Morris