Anthropic PBC has unveiled Claude Mythos, its most advanced AI model, designed to enhance software security.
The model, part of Project Glasswing, is being shared with over 40 cybersecurity partners including Amazon, Apple, Microsoft, and Cisco.
Mythos was not originally built for security but demonstrates exceptional ability in detecting flaws in code-such as a 16-year-old flaw in FFmpeg.
It’s capable of identifying thousands of vulnerabilities across major platforms like Windows, macOS, and Android.
According to Anthropic, Mythos can also generate exploits instantly-a dual-edged sword. However, it offers immense potential for improving defense mechanisms.
The company plans to scale Mythos in the future but must first implement robust safeguards against misuse.