Autonomous AI Agent Executes Full Ransomware Attack Without Human Intervention

Security researchers at Sysdig documented JadePuffer, an AI agent that conducted an entire ransomware attack autonomously. The agent exploited a Langflow vulnerability, accessed a MySQL database, escalated privileges, encrypted over 1,300 records, and generated its own ransom note. Human involvement was limited to infrastructure setup and target selection, compressing what normally requires weeks of manual work into a single operation.