The “first” fully agentic ransomware is here, but we aren’t panicking (yet): Meet JadePuffer

Sysdig research and reporting describe the JadePuffer campaign as an LLM-driven ransomware attack that exploited CVE-2025-3248 and used an autonomous agent to retry failed steps, but notes the attack was directed by an apparently unskilled human operator (no stored decryption keys or extortion address); the key takeaway is that responsibility and mitigation still hinge on human actors and basic security fundamentals.