AI Breaches the World's Most Secure OS in Under 10 Hours, Elite Hacker Team Would Need Weeks

Artificial Intelligence (AI) technology has recently showcased its prowess again, potentially threatening future digital security. This time, AI was used to hack the world’s most secure operating system (OS), FreeBSD, and succeeded.

Not only did it succeed, but the breach of FreeBSD by AI occurred remarkably quickly. The AI successfully exploited a critical security vulnerability in the FreeBSD kernel (code CVE-2026-4747) remotely in less than 10 hours, specifically 4 to 8 hours.

News of this high-level hacking carried out by AI was revealed by cybersecurity expert Amir Husain in an analysis report published in Forbes. In his report, Husain explained that the AI model was built using Claude from Anthropic.

The AI independently designed a complete remote code execution (RCE) attack chain. The code then successfully obtained root shell access, the highest level that allows full control of the server system.

Throughout the hacking process, the AI was able to design its own testing environment using the QEMU emulator.

It even constructed a complex return-oriented programming (ROP) chain and was capable of identifying issues and performing independent debugging when its exploitation failed or got stuck midway.

“For those involved in cybersecurity, this is a crucial threshold moment. We have moved from an era where AI was merely a tool for security researchers to becoming a fully autonomous actor capable of conducting sophisticated offensive operations against systems,” wrote Husain, as quoted by KompasTekno from Forbes.

This case shows that cyber attacks are undergoing a transformation. In the past, high-level cyber attacks required very expensive costs and the availability of human hackers with decades of experience.

Now, AI can do it with very low costs and quickly.

Facing this new era, large companies and national security institutions are urged to immediately overhaul their cyber defence strategies entirely.

In the future, cyber battles seem no longer to occur between humans versus humans, but rather artificial intelligences hacking each other.