Anthropic's Project Glasswing Utilizes AI to Enhance Cybersecurity by Identifying Software Vulnerabilities
Anthropic, an AI firm, has launched Project Glasswing, an initiative that leverages artificial intelligence to identify and fix cybersecurity vulnerabilities in critical software. The project utilizes Claude Mythos Preview, a powerful AI model capable of understanding and modifying complex software to autonomously detect and remediate vulnerabilities. During testing, the model identified numerous zero-day vulnerabilities, including a 27-year-old flaw in OpenBSD and a 16-year-old issue in FFmpeg. The project has garnered support from major tech companies like Amazon, Apple, and Google, which have tested the model's capabilities. Anthropic aims to provide up to $100 million in usage credits to organizations maintaining critical software infrastructure. 
