JADEPUFFER: Security Researchers Document the First AI Agent to Run a Full Ransomware Campaign Autonomously
Sysdig has documented JADEPUFFER — an LLM agent that autonomously ran a complete ransomware campaign across 600+ payloads, with no human directing each step. It's the first confirmed case of its kind. »
Kali Linux 2026.2: Nine New Tools, GNOME 50, and VM Boot Time Cut by Two-Thirds
Kali Linux 2026.2 arrives with nine new security tools including a multiprotocol bruteforcer and an LLM command assistant, GNOME 50, KDE Plasma 6.6, and a VM boot time cut to one third thanks to trimmed initrd images. »
The Linux Foundation Just United 19 Tech Giants to Shield Open Source From AI-Powered Attacks
Launched June 25, Akrites is a new coordinated vulnerability-response initiative backed by AWS, Anthropic, Google, Microsoft, NVIDIA, and 14 other organizations committed to defending the open source software the world runs on. »
Anthropic Accuses Alibaba of Stealing Claude's Intelligence With 25,000 Fake Accounts
In the largest known distillation attack against a U.S. AI company, Anthropic says Alibaba ran 28.8 million exchanges through fraudulent Claude accounts to train its Qwen models. »
Anthropic's AI Found 10,000 Security Holes — Now It's Scanning Critical Infrastructure
Anthropic expanded Project Glasswing to 150 organizations across 15 countries, deploying Claude Mythos Preview to find and fix critical vulnerabilities in hospitals, energy grids, and telecom networks. »