Microsoft has introduced its first artificial intelligence model built specifically for cybersecurity. The company also launched a new AI-powered platform designed to automate security operations and help organizations detect and respond to cyber threats more efficiently.
The new model, MAI-Cyber-1-Flash, focuses on identifying complex software vulnerabilities before attackers can exploit them. Microsoft has integrated the model into its MDASH system, which coordinates multiple AI models and more than 100 specialized AI agents to identify, verify, and fix software security issues.
The company also announced Project Perception, a platform that uses AI agents to monitor systems, investigate suspicious activity, and respond to security incidents in real time.
MAI-Cyber-1-Flash Improves Vulnerability Detection
Microsoft designed MAI-Cyber-1-Flash to scan large and complex codebases for hidden security weaknesses. Instead of working alone, the model operates within MDASH, which assigns cybersecurity tasks to the most suitable AI model or agent.
According to Microsoft, MAI-Cyber-1-Flash can complete up to 90 percent of MDASH’s security tasks. Meanwhile, larger AI models such as GPT-5.4 handle the most demanding workloads. This approach helps reduce scanning costs while improving efficiency.
Microsoft reported that MDASH, powered by MAI-Cyber-1-Flash and GPT-5.4, achieved a score of 95.95 percent on the CyberGym benchmark. The company said this performance exceeded competing AI systems while reducing operating costs by nearly 50 percent compared with its previous setup.
Project Perception Uses AI Teams to Respond to Threats
Project Perception extends beyond vulnerability detection by using AI agents to manage cybersecurity operations continuously. Microsoft divides these agents into red, blue, and green teams.
Red-team agents search for weaknesses from an attacker’s perspective. Blue-team agents investigate suspicious activity and assess security risks. Green-team agents then strengthen protections and help resolve identified issues.
Microsoft said the platform can identify vulnerabilities, evaluate risks, and recommend or generate code fixes within minutes. Previously, these tasks often required hours of manual work by cybersecurity specialists.
Microsoft Expands AI Cybersecurity Capabilities
Microsoft said cybercriminals increasingly use artificial intelligence to discover software weaknesses and launch attacks more quickly. As a result, organizations need security tools that operate at a similar speed.
The company stated that its security systems process more than 100 trillion security signals every day across devices, cloud services, networks, applications, and digital identities.
Microsoft joins a growing market of AI cybersecurity providers alongside Anthropic and OpenAI. Project Perception will enter public preview on August 3, while MAI-Cyber-1-Flash is already being deployed through MDASH to strengthen cybersecurity operations.
