Microsoft has developed an advanced AI system that can reverse-engineer and identify malicious software without any human help. Named Project Ire, this prototype system automatically dissects software files to figure out what they do and if they're dangerous—a process typically performed by human security experts, the company said.

Essentially, Project Ire's approach is a departure from existing security tools that scan for known threats. According to Microsoft, the AI "automates what is considered the gold standard in malware classification: fully reverse engineering a software file without any clues about its origin or purpose." This deep, autonomous analysis is crucial as both security defenders and hackers increasingly use AI to their advantage, according to the company.

The company said that in a recent test, Project Ire demonstrated its accuracy by creating a threat report that was strong enough to automatically block an advanced piece of malware. According to Microsoft, early tests show the AI is highly accurate. When the system identified a file as malicious, it was correct 98% of the time, with a false positive rate of just 2%.

The development of Project Ire is part of a broader company-wide push on security. Following a series of high-profile vulnerabilities, Microsoft has made security its top priority through its Secure Future Initiative. Like Google's "Big Sleep" AI, which focuses on discovering vulnerabilities in code, Project Ire is part of a new wave of AI systems designed to address cybersecurity threats in novel ways. The system, developed by teams across Microsoft Research, Microsoft Defender, and Microsoft Discovery & Quantum, will now be used internally to help speed up threat detection across Microsoft's own security tools.