Arabic version: قادة المالية يعبرون عن القلق بشأن مخاطر الأمان لنموذج الذكاء الاصطناعي الجديد
Finance ministers and top bankers have voiced significant concerns regarding the Claude Mythos AI model developed by Anthropic, fearing it may compromise the security of financial systems. This urgency led to crisis meetings following the model’s identification of vulnerabilities in major operating systems. According to BBC News, Canadian Finance Minister François-Philippe Champagne emphasized the need for safeguards, stating that the model presents unknown risks that require immediate attention to ensure financial resiliency.
Mythos, part of Anthropic’s Claude AI system, has been highlighted for its ability to identify and exploit cybersecurity weaknesses, prompting extensive discussions at the International Monetary Fund (IMF) meeting in Washington, DC this week. Experts have noted that while the model shows potential in computer security tasks, further testing is necessary to fully understand its capabilities. Anthropic has opted not to publicly release Mythos, instead providing access to major tech companies like Amazon and Microsoft as part of Project Glasswing, aimed at securing critical software.
Concerns surrounding Mythos have sparked debates among cybersecurity professionals, with some questioning the justification for the alarm given the lack of broader industry testing. A report from the UK’s AI Security Institute indicated that while Mythos is a powerful tool for identifying security flaws, it does not significantly outperform its predecessor, Claude Opus 4.
Top bankers, including Barclays CEO CS Venkatakrishnan, have been granted early access to the model to assess its implications for their systems. Bank of England Governor Andrew Bailey highlighted the seriousness of the situation, noting that the development of such AI tools could facilitate cybercriminal activities by exposing vulnerabilities in core IT systems. The U.S. Treasury has also urged major banks to evaluate their systems proactively before Mythos’s public release, reflecting a growing consensus on the need for vigilance in the evolving landscape of AI and cybersecurity.
