JadePuffer: World’s ‘first AI ransomware’ capable of autonomous attacks

Artificial intelligence (AI) has now reached a pinnacle in terms of cybersecurity, though not through server reinforcement but server destruction.

Meet JadePuffer, the first AI-driven agentic ransomware, meaning it carries out attacks with no human intervention when it comes to technical execution.

Unlike traditional ransomware that follows pre-programmed instructions, JadePuffer can analyze its target and find the optimal way to spread or encrypt files. 

According to researchers from the company of Sysdig, JadePuffer was tested against an internet-exposed server and configuration service, with the AI successfully operating its schemes. 

The model enumerated the database contents, searched for important data, deleted the database then left an extortion note for money. 

With this, it highlights the rapid pace of technological evolution among AI models, even executing cyberattacks meant for ransom. It proves the adaptability of AI in the modern world, and how it can be used as a tool for both convenience and harm.

However, although the AI can operate without human intervention for its technical execution, the company noted that a human would still need to set up the AI before it attacks.