Microsoft has introduced its first AI model built specifically for cybersecurity alongside Project Perception, a platform that coordinates teams of agents to identify, assess and fix security weaknesses. Microsoft AI CEO Mustafa Suleyman said the system performed better than Gemini, GPT-5.5 Cyber, GPT-5.6 Sol and Anthropic’s Mythos 5. Microsoft said the platform can combine information from identities, devices, applications, cloud services, data and AI systems. AI Security Market Becomes More CompetitiveMicrosoft said the new tools are intended to help organisations respond as cybercriminals increasingly use AI to automate attacks. Microsoft’s wider cybersecurity products and research are available through the company’s Security Blog and Microsoft Security website.