川普政府對Anthropic祭制裁 美法官裁暫緩執行 | 國際 | 中央社 CNA
2026-03-27
Central News Agency
The information displayed in the AIM should not be reported as representing the official views of the OECD or of its member countries.
A major data leak exposed details of Anthropic's powerful new AI model, Claude Mythos/Capybara, revealing advanced cybersecurity exploitation capabilities. The leak, caused by human error, led to real-world misuse attempts by hacking groups and triggered a sharp decline in cybersecurity stocks, highlighting significant AI-driven cybersecurity risks.[AI generated]
Why's our monitor labelling this an incident or hazard?
The event involves an AI system (Mythos AI model) whose development details were leaked due to a human error in system configuration. While no actual harm has been reported, the model's advanced capabilities, especially in cybersecurity and coding, present a credible risk of misuse or malicious use that could lead to harm in the future. The leak itself does not constitute an incident since no harm has occurred, but the potential for harm is significant, making this an AI Hazard. The article focuses on the leak and the model's capabilities rather than any realized harm or ongoing incident, so it does not qualify as an AI Incident or Complementary Information.[AI generated]