Self-Evolving Defense: Continual Security Policy Learning for LLM Agents

Large language models LLMs increasingly power agents that access sensitive information, use external tools, and modify software repositories. Although these capabilities offer substantial benefits, they also create security risks such as jailbreaks, prompt injection, and vulnerable code generatio...

Surgical Repair of Insecure Code Generation in LLMs

Large language models write production code, and yet they routinely introduce well-known vulnerabilities. We show that this is not a knowledge deficit: the same models that generate insecure code, correctly identify and explain the vulnerability when asked directly, this is a gap we call the...

Minecraft-Rcon-Manage security vulnerability

Minecraft-Rcon-Manage is a game server framework developed by MemoryCache’s individual developers. Versions of Minecraft-Rcon-Manage prior to version 3 contained security vulnerabilities, which stemmed from improper control over code generation, potentially leading to code injection attacks...

Mind the Gap: a Practical Attack on GGUF Quantization

With the increasing size of frontier LLMs, post-training quantization has become the standard for memory-efficient deployment. Recent work has shown that basic rounding-based quantization schemes pose security risks, as they can be exploited to inject malicious behaviors into quantized models tha...

JumpServer WEB CLI Arbitrary Command Execution

The vulnerability of the WEB CLI interface component koko of the JumpServer security audit and maintenance system is related to incorrect code generation. Exploiting this vulnerability allows a remote attacker to execute arbitrary commands...

ChatGPT writes insecure code

Research by computer scientists associated with the Universite du Quebec in Canada has found that ChatGPT, OpenAI's popular chatbot, is prone to generating insecure code. "How Secure is Code Generated by ChatGPT?" is the work of Raphael Khoury, Anderson Avila, Jacob Brunelle, and Baba Mamadou...

flex: Potential insecure code generation

Background flex is a programming tool used to generate scanners programs which recognize lexical patterns in text. Description Chris Moore discovered a buffer overflow in a special class of lexicographical scanners generated by flex. Only scanners generated by grammars which use either REJECT, or...