AI Agent Exploits Langflow RCE to Automate Database Ransomware Attack
The security company Sysdig claims to have discovered what it thinks to be the first ransomware attack carried out entirely by an AI bot.
The operator is referred to as JADEPUFFER by its Threat Research Team, which claims that a huge language model performed the entire task, including breaking in, collecting credentials, penetrating farther into the network, encrypting, and erasing a company's production database.
Ransomware has always required a trained individual to be involved, either at the keyboard or in writing the script that the infection uses. The skill required to launch an attack decreases to the cost of hiring an AI agent if a model is able to chain those actions on its own.
An old bug that had already been fixed was the entry point. JADEPUFFER took use of CVE-2025-32...