Anthropic's Mythos AI: New Name, Stronger Safety Controls

Anthropic, the AI company known for its Claude models, is reportedly preparing to release its advanced Mythos AI model to the public. It may launch this model under a new name and with stronger safety measures. The company plans to apply stricter usage controls to reduce risks associated with misuse and highly advanced outputs. Mythos was not initially released publicly due to concerns it could be misused for cyberattacks, given its strong ability to analyze software flaws. Instead, Anthropic tested Mythos through Project Glasswing, providing secure access to selected organizations. This project expanded to over 15 countries and included more than 100 organizations, with companies like Apple, Google, and Microsoft participating. Mythos identified thousands of serious vulnerabilities in widely used software. This matters because AI systems like Mythos could be used by attackers to exploit unpatched vulnerabilities.