Anthropic's New AI, Mythos, Is a 'Cyber Reckoning'—But It's Too Dangerous to Release

The AI company Anthropic announced a powerful new artificial intelligence model named Mythos on Tuesday. It calls the technology a potential "reckoning" for cybersecurity. However, the company stated it is too dangerous to release publicly. Instead, Anthropic is working with 40 security firms and corporations to test how Mythos can be used to prevent cyberattacks. The goal is to see if the AI can help defenders find and fix software vulnerabilities before hackers can exploit them. This controlled testing aims to ensure the powerful tool is used for protection, not for creating new cyber threats.