Anthropic eyes public Claude Mythos rollout

The transition of Mythos from a "too dangerous to release" research project to an enterprise-ready security tool marks the beginning of the AI-native cybersecurity era.

• –10,000+ high-severity vulnerabilities found in 30 days proves that AI-driven offensive security is no longer a theoretical threat but a current reality.
• –Mythos reportedly identified flaws in every major operating system and browser, including a 27-year-old OpenBSD bug that human researchers missed for decades.
• –The primary bottleneck has officially shifted from bug discovery to patching; human maintainers are currently overwhelmed by the volume of AI-generated security disclosures.
• –Integration into Claude Code suggests Anthropic is "shifting left," allowing developers to fix vulnerabilities at the point of creation rather than after deployment.
• –Pricing remains a significant barrier, with input/output tokens costing roughly 5x more than Claude 4.7 Opus.