Microsoft Empowers AI To Combat Computer Viruses

Microsoft engineers have shared details about the development of a cutting-edge artificial intelligence (AI) agent designed to analyze and detect malware with impressive accuracy, Azernews reports.

The prototype, named Project Ire, is an advanced system trained to reverse-engineer malicious software, identify threats, and even help neutralize computer viruses - tasks typically reserved for experienced cybersecurity specialists.

During the malware analysis process, the AI agent determines the file type, dissects its structure, and pinpoints potentially malicious components. One of the standout features of Project Ire is its ability to generate a "chain of evidence" for each decision it makes, offering transparency and insight into the agent's logic - something that's often lacking in black-box AI systems.

In internal testing, Project Ire demonstrated promising results: it successfully detected 90% of malicious Windows driver files, while maintaining a low false-positive rate of just 2%. According to Microsoft, this level of precision highlights the system's potential for integration into real-world security operations.

“Such a low level of false positives suggests a clear potential for using the system in security operations,” Microsoft stated.

The company believes that AI agents like Project Ire could significantly ease the workload of cybersecurity teams, especially during large-scale cyberattacks, when speed and efficiency are critical.

However, Microsoft also acknowledges that the agent is still a work in progress. While the current results are encouraging, the system's accuracy and threat classification capabilities are not yet on par with human analysts in every scenario. The company describes Project Ire's current performance as "moderate", and further improvements are underway.

What sets Project Ire apart is its potential to become part of a broader automated cyber defense ecosystem, where AI doesn't just detect threats - it responds to them autonomously. Imagine future security systems that learn from each cyberattack, adapting in real time, and even coordinating with other AI agents across networks to defend against sophisticated malware campaigns. Project Ire might just be a glimpse into that future.

MENAFN08082025000195011045ID1109901915