General Analysis Launches MCP Guard
•3 min read
We are excited to launch MCP Guard, the first runtime firewall designed to secure every MCP (Model Context Protocol) tool call against prompt injection attacks.
Blog
The Redact-and-Recover Jailbreak Reveals Ricin Extraction Instructions from Claude
•8 min read
We present the Redact & Recover (RnR) Jailbreak, a novel attack that exploits partial compliance behaviors in frontier LLMs to bypass safety guardrails through a two-phase decomposition strategy.
Supabase MCP can leak your entire SQL database
•8 min read
In this post, we show how an attacker can exploit Supabase’s MCP integration to leak a developer’s private SQL tables. Model Context Protocol (MCP) has emerged as a standard way for LLMs to interact with external tools. While this unlocks new capabilities, it also introduces new risk surfaces.
General Analysis x Together AI
•2 min read
TLDR: We are excited to announce our partnership with Together AI to stress-test the safety of open-source (and closed) language models.
The Jailbreak Cookbook
•40 min read
We have created a comprehensive overview of the most influential LLM jailbreaking methods.
Generating Diverse Test Cases with Diversity Transfer from LegalBench
•5 min read
TLDR: we utilized LegalBench as a diversity source to enhance the diversity of our generation of red teaming questions. We show that diversity transfer from a domain-specific knowledge base is a simple and practical way to build a solid red teaming benchmark.
Red Teaming GPT-4o: Uncovering Hallucinations in Legal AI Models
•5 min read
In this work we explore automated red teaming, applied to GPT-4o in the legal domain. Using a Llama3 8B model as an attacker, we generate more than 50,000 adversarial questions that cause GPT-4o to hallucinate responses in over 35% of cases.