New research reveals that DeepSeek-R1 produces more security vulnerabilities in code generated from prompts containing politically sensitive topics for China, such as Tibet or Uyghurs.
Replays of the .conf25 Global Broadcast sessions, including the Welcome Keynote, Product Keynote, and various sessions covering topics like AI, security, observability, and Splunk platform updates.
Fly.io provides a secure and fast platform for deploying AI workflows and LLM-generated code using ephemeral, kernel-isolated virtual machines (Fly Machines). It offers features like secure sandboxing, fast startup times, a clean slate for each run, a simple API, and support for whole applications, not just code snippets.
This week's security roundup covers the Anubis web AI firewall, AI exploit generation, a vulnerability in CodeRabbit, the potential illegality of adblocking in Germany, a Microsoft Copilot audit log issue, and a disputed Elastic EDR vulnerability.
Trail of Bits announces the open-sourcing of Buttercup, their AI-driven Cyber Reasoning System (CRS) developed for DARPA’s AI Cyber Challenge (AIxCC). The article details how Buttercup works, including its four main components (Orchestration/UI, Vulnerability discovery, Contextual analysis, and Patch generation), provides instructions for getting started, and outlines future development plans.
This article details significant security vulnerabilities found in the Model Context Protocol (MCP) ecosystem, a standardized interface for AI agents. It outlines six critical attack vectors – OAuth vulnerabilities, command injection, unrestricted network access, file system exposure, tool poisoning, and secret exposure – and explains how Docker MCP Toolkit provides enterprise-grade protection against these threats.
This article details the Model Context Protocol (MCP), an open standard for connecting AI agents to tools and data across enterprise landscapes. It covers MCP implementations by AWS, Azure, and Google Cloud, security considerations, and the growing ecosystem surrounding the protocol.
A review of a Google paper outlining their framework for secure AI agents, focusing on risks like rogue actions and sensitive data disclosure, and their three core principles: well-defined human controllers, limited agent powers, and observable actions/planning.
This article discusses a new paper outlining design patterns for mitigating prompt injection attacks in LLM agents. It details six patterns – Action-Selector, Plan-Then-Execute, LLM Map-Reduce, Dual LLM, Code-Then-Execute, and Context-Minimization – and emphasizes the need for trade-offs between agent utility and security by limiting the ability of agents to perform arbitrary tasks.
Cisco and Meta are championing open-source large language models (LLMs) for enterprise threat defense, announcing new models and initiatives at RSAC 2025. Cisco's Foundation-sec-8B LLM and Meta's AI Defenders Suite aim to provide scalable, secure, and cost-effective cybersecurity solutions through collaboration and open innovation.
