An AI model from Anthropic introduced in 2026 that can detect vulnerabilities in software code. Mythos has been able to find issues in most all common software, and it has been released to software vendors rather than the general public.
Although other AI models have been known to expose vulnerable software, Claude Mythos apparently does a superior job, which caused a lot of anxiety.
Project Glasswing
Anthropic's Project Glasswing was designed to secure all critical software in use today using Mythos. Glasswing partners include companies such as Apple, Microsoft, Amazon, NVIDIA and Google.
In the Wrong Hands?
If Mythos is deployed by hackers, it can be used to corrupt most of the operating systems and applications in use today before they are patched by their vendors. See
Anthropic.