Microsoft is developing a powerful new AI tool that it says could meet the “gold standard” for detecting and analyzing malware. The new tool, called Project Ire, is still a prototype, but it has already shown it can identify malicious files even without knowing anything about them beforehand.
The ultimate goal is to build Project Ire directly into Microsoft Defender, the company’s built-in antivirus software. It would act as a “Binary Analyzer,” capable of spotting malware from any source the first time it encounters it.
So how well does it work? In Microsoft’s tests, Project Ire correctly identified nearly 9 out of 10 malicious files, a very promising start. However, it only managed to catch about a quarter of all the malware in the test set, and it had a false positive rate of 4%, meaning it sometimes flagged clean files as malicious.
Despite the mixed results, Microsoft is optimistic. “While overall performance was moderate, this combination of accuracy and a low error rate suggests real potential for future deployment,” the company said. In a separate test, the tool performed even better, with a higher detection rate and a lower false positive rate.
Microsoft plans to keep improving the tool’s speed and accuracy, with the hope of one day adding it to Microsoft Defender. As cybercriminals increasingly use AI to create new threats, this project shows how cybersecurity experts are using the same technology to fight back.
