ANTHROPIC has unveiled Claude Mythos, a new AI model designed to strengthen cybersecurity through Project Glasswing, a collaborative effort to defend critical software before hackers can exploit it. The project, led by Anthropic and supported by partners including Amazon Web Services, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks, aims to detect and fix vulnerabilities at scale using Mythos Preview.
Anthropic says Mythos is a major leap beyond its previous models, with a new top tier called Copybara and strong agentic coding and reasoning capabilities that excel at software tasks. The initiative also seeks to limit access for now and to improve global cybersecurity by funding usage credits and open-source security projects to enable wider collaboration.
The piece notes that global cybercrime is estimated at around $500 billion annually and warns that AI-driven capabilities could accelerate attacks, underscoring the urgency to deploy defensive tools like Glasswing, said to help identify and fix flaws before exploitation. according to Cisco.