Microsoft has unveiled a new artificial intelligence system that can independently detect and block malware without any human assistance. Called the project ire, it is designed for reverse-engineer software files to the prototype agent and is determined whether they are safe or harmful, marking a major step in cyber security. According to Microsoft’s blog post, the project ire can fully analyze a software file, even if it does not have any prior information about the source or purpose of the file. It uses dicampillars and other advanced devices to scan the code, understand its behavior and decide whether it causes risk. This tool is the result of a joint effort between Microsoft Research, Microsoft Defender Research and Microsoft Discovery and Quantum.

“This type of work has been traditionally done by expert analysts, which can be slow and tiring,” Microsoft explained. Security researchers often suffer from cautious fatigue and burnouts, making it difficult to maintain continuity in large -scale malware detection.

Project Ire stands out of other AI safety devices because the malware classification is particularly difficult to automate. There is no clear way to verify your decisions for the machine, and many symptoms of malicious software may also appear in legitimate programs. This makes it difficult to train a system that is both accurate and reliable.

To deal with this, Microsoft equipped the project ire with a system that calls it a “series of evidence”, a step-by-step trace shows how the agent reached its conclusion. This audit trail allows human experts to later verify their findings and improve accountability in case of errors.

The analysis of the project IRE begins with the file type and structure tringing, then rebuilding its control flow using devices such as ghidara and ARGR. This can then call separate tools through API, which can summarize each code function, add the results to its evidence series.

Microsoft tested the agent in two major assessments. In a test, it analyzed a dataset of Windows drivers, some malicious, others safe. AI correctly identified 90 percent of the files, in which only 2 percent of safe files were incorrectly flagged as danger. This reminded the project to be an exact score of 0.98 and 0.83.

In a difficult real -world testing, Microsoft gave AI around 4,000 complex files, which were not yet reviewed by any other automated systems. These files were for manual inspection by experts. Even under these conditions, Project Ire achieved a high precision score of 0.89, which had a false positive rate of only 4 percent.

In fact, the project Ire Microsoft had the first reverse engineer, human or machine, which was strong enough to produce a malware detection case to justify the automatic blocking of an advanced consistent danger (APT) sample. That malware has now been neutralized by Microsoft defender.