Microsoft has unveiled a new AI cybersecurity system, Project Perception, which utilizes red, blue, and green agents to identify attack paths, evaluate risks, and rectify vulnerabilities. This innovative system is powered by models from Microsoft, OpenAI, and Anthropic. The public preview of the platform is set to begin on August 3, and Microsoft reports that Project Perception achieved a score of 96% on CyberGym, demonstrating its effectiveness in enhancing cybersecurity measures.

OpenAI: OpenAI is an artificial intelligence research and deployment organization focused on building advanced generative models for applications in productivity, analysis, and security. Its models are incorporated into Microsoft’s new cybersecurity system to support agent-based risk assessment and flaw remediation processes.
Anthropic: Anthropic is an AI company dedicated to developing safe and reliable large language models for enterprise and research use. Its models contribute to the agent framework in Microsoft’s cybersecurity platform, aiding in the identification of threats and automated responses.
Microsoft: Microsoft is a multinational technology company that develops software, cloud computing services, and artificial intelligence solutions used globally across enterprises and consumers. The company has unveiled a new AI cybersecurity system that deploys specialized agents to detect attack paths, evaluate risks, and remediate issues. This platform integrates models from OpenAI and Anthropic to enhance its defensive capabilities.

Availability: Public preview of the cybersecurity platform begins on August 3.
Product Launch: Microsoft is releasing a new AI cybersecurity system that uses red, blue, and green agents powered by models from multiple providers.
Integration Approach: The system combines Microsoft’s own models with those from OpenAI and Anthropic to address attack paths, risk evaluation, and vulnerability fixes.