According to the Financial Times, UK financial regulators are holding urgent discussions with the National Cyber Security Centre and several banks to assess the potential risks posed by the latest large model launched by AI company Anthropic, "Claude Mythos Preview." The model was found to have thousands of critical software vulnerabilities in the "Glass Wing Project."

According to insiders, officials from the Bank of England, the Financial Conduct Authority, and the Treasury are working closely with the National Cyber Security Centre to confirm the potential vulnerabilities of the "Claude Mythos Preview" model on critical information systems. Representatives from major UK banks, insurance companies, and exchanges will attend meetings with regulatory authorities in the next two weeks to receive specialized briefings on the cybersecurity risks of this model.

Meanwhile, US Treasury Secretary Scott Bessent has also convened a meeting with major Wall Street banks to discuss the potential cybersecurity risks posed by this model. Anthropic stated that the model is used under the controlled "Glass Wing Project," accessible only to specific institutions for cybersecurity defense purposes.

Notably, Anthropic mentioned in a recent blog post that the model has identified thousands of critical vulnerabilities in widely used software such as operating systems and web browsers, further drawing the attention of regulatory authorities.

With the rapid development of artificial intelligence technology, ensuring its security has become an important challenge for the global financial system. The regulatory actions taken by the UK reflect a high level of concern about the potential risks of new technologies and aim to safeguard the security and stability of the financial system through timely assessments and communication.

Key Points:

🌐 UK financial regulators are urgently discussing with the cybersecurity center and banks to assess the potential risks of the AI model.  

🔍 The "Claude Mythos Preview" model was found to have thousands of critical software vulnerabilities.  

📅 Meetings will be held in the next two weeks, with representatives from banks and insurance companies listening to cybersecurity risk briefings.