Anthropic Unveils Mythos Model for Cybersecurity with Partner Organizations
Anthropic has launched a preview of its Mythos model, designed for cybersecurity applications, in collaboration with over 40 partner organizations. Mythos, part of the Project Glasswing initiative, is being used to identify vulnerabilities in both internal and public systems. Despite not being specifically trained for cybersecurity, Mythos has already detected thousands of critical zero-day vulnerabilities. Partners such as Amazon, Apple, and Microsoft are involved in testing the model, which is not yet publicly available. Anthropic is also in discussions with federal officials regarding the model's use, amidst legal disputes with the US government. 
