A Russian-speaking access broker scanned 430,000 FortiGate firewalls, deployed custom packet sniffers to harvest 110M+ credentials, and fueled ransomware operations. Here's what IT teams need to know — and do — right now.
OpenAI's GPT-5.6 launches with three models — Sol, Terra, and Luna — delivering state-of-the-art coding and cybersecurity results at roughly one-third the cost of the previous frontier tier. Ultra mode coordinates four parallel agents for complex tasks.
JADEPUFFER is the first documented AI-driven ransomware operation with no human at the keyboard. An LLM agent exploited Langflow, stole credentials across 6 categories, pivoted to production, corrected its own errors in 31 seconds, and encrypted 1,342 records autonomously.
A decades-old Unix trick broke six AI coding assistants in one week, and an autonomous AI agent ran its first complete ransomware attack. Here's what developers and security teams need to understand — and fix.