- Microsoft introduced its first AI model built specifically to find security flaws in source code.
- The model, called MAI-Cyber-1-Flash, will work with OpenAI's GPT-5.4 inside Project Perception, a group of AI agents designed to detect and repair vulnerabilities.
- Project Perception will enter public preview on Aug. 3.
- Microsoft said the combined system outperformed competing security models on the CyberGym benchmark while operating at about half the cost.
- The launch is Microsoft's first major cybersecurity initiative since Hayete Gallot returned in February to lead the business.
- Microsoft has not updated the size of its cybersecurity business since 2023, when annual revenue topped $20 billion.
- AI-Cyber-1-Flash is not so much active malware protection as it is a software vulnerability analysis tool, and if anyone knows software vulnerability it’s Microsoft.
- The model is training on Microsoft’s decades of patching and analysis as well as processing the roughly 1 trillion security signals triggered by their 1.6 million customers every day.
- The other announcement is a collection of AI-powered agents called Project Perception and is intended to automate at least some of the five steps of the NIST cybersecurity framework.
- Microsoft expects Project Perception to automate a fair amount of cybersecurity tasks.
- "As AI accelerates the speed and scale of cyberattacks, defenders are being asked to secure increasingly complex digital environments with approaches built for a different era," the company said.
Microsoft has unveiled MAI-Cyber-1-Flash, its inaugural AI model tailored for cybersecurity, designed to identify security flaws in source code. This model integrates with OpenAI's GPT-5.4 within Project Perception, a suite of AI agents aimed at automating vulnerability detection and repair.1239
The launch marks a significant step for Microsoft, being its first major cybersecurity initiative since Hayete Gallot took the helm in February. The company has not disclosed the current size of its cybersecurity business, which surpassed $20 billion in annual revenue in 2023. Microsoft has a rich history in cybersecurity, having transformed Windows Defender from a criticized product into a formidable competitor in malware protection.56

MAI-Cyber-1-Flash is primarily a software vulnerability analysis tool, leveraging Microsoft's extensive experience in patching and analyzing security threats. The model processes approximately 1 trillion security signals generated daily by its 1.6 million customers. Microsoft claims that the new system outperformed existing security models on the CyberGym benchmark while operating at about 50% of the cost of its competitors.478
As cyberattacks grow in speed and complexity, Microsoft emphasizes the need for modern solutions, stating, "As AI accelerates the speed and scale of cyberattacks, defenders are being asked to secure increasingly complex digital environments with approaches built for a different era," highlighting the urgency for innovative cybersecurity measures.11
“The model is set to work with OpenAI's GPT-5.4 and is part of Project Perception, which aims to automate steps of the NIST cybersecurity framework. Microsoft claims the combined system outperformed competitors on the CyberGym benchmark while operating at about half the cost.”
