News · moment 15 of 20
Anthropic's Too-Dangerous-to-Release AI Model
from Bombing Beirut, Back To Butler & Melania’s Mission | The Tim Dillon Show #491 · #491
2 min
News
Tim describes Anthropic restricting access to its Mythos AI model after it proved extraordinarily good at finding critical security vulnerabilities, and details the Project Glasswing arrangement with major tech companies.
- Anthropic develops an AI tool deemed too dangerous to release to the public because it can identify weaknesses in computer systems.
- The company restricted access to its Claude Mythos preview model after discovering it was so effective at finding critical security flaws it could reshape cybersecurity or cause havoc in wrong hands.
- The system has already discovered thousands of security vulnerabilities in popular web browsers and operating systems.