Microsoft announced plans Wednesday to integrate advanced artificial intelligence models, including Anthropic's Claude Mythos Preview, into its secure coding framework. This move aims to significantly enhance the company's cybersecurity capabilities.
By incorporating these AI models into Microsoft's Security Development Lifecycle (SDL), the company expects to identify vulnerabilities and develop fixes at an earlier stage of development. Mythos, revealed on April 7, has already identified thousands of critical vulnerabilities in various software types. Its advanced coding abilities allow it to pinpoint cybersecurity weaknesses and devise potential exploitation methods.
Anthropic is deploying Claude Mythos Preview to select companies through its "Project Glasswing" initiative. This controlled rollout allows major tech firms, including Microsoft, Amazon.com, and Apple, to leverage the AI for cybersecurity vulnerability searches.
Microsoft reported substantial improvements after evaluating Mythos using its internal open-source benchmark for real-world detection engineering tasks. The AI's performance surpassed prior models in identifying vulnerabilities.
The integration comes as global institutions, including the U.S. administration and central banks, are actively assessing AI's impact on cybersecurity, recognizing its potential to both facilitate and counter complex cyberattacks.