Feared by Finance Minister and Bankers: The Capabilities of Anthropic's Mythos AI

The name “Claude Mythos” has suddenly become a serious topic of discussion among world economic leaders. Anthropic’s latest artificial intelligence (AI) model is not just an ordinary technological innovation. It has triggered deep concerns among finance ministers, bankers, and global financial industry players, who assess that its capabilities could shake the stability of the world’s financial system. These concerns have even echoed in prestigious international forums, including the International Monetary Fund (IMF) meeting in Washington DC last April, where Mythos became one of the most crucial topics debated by world economic leaders. So, what exactly is Claude Mythos, and what capabilities make it so feared? Here is the review. Anthropic announced the existence of Mythos on 7 April, but emphasised that the model will not be released to the public due to its ability to identify hidden gaps in IT systems that could be exploited by hackers. Most worryingly, Mythos is claimed to be capable of identifying and exploiting “zero-day” gaps, namely weaknesses that are completely unknown to developers or affected organisations, in various important operating systems and browsers. Some gaps found have even been reported to have existed for decades without ever being detected. Anthropic calls this a “historic moment for cybersecurity”. The UK’s AI Security Institute (AISI) has evaluated Mythos and concluded that the model represents a “significant improvement” compared to previous models. Mythos successfully completed a 32-step cyber attack simulation in a test designed by AISI, a first achievement that no previous AI model has accomplished. Nevertheless, several experts warn against getting swept up in panic. An AI security company named Aisle found that far cheaper models can also identify similar gaps. Experts also remind that most security breaches still stem from conventional weaknesses such as weak authentication and known gaps that have not yet been patched. Not a few parties assess that there is an element of hype in the way Anthropic, a startup valued at around $800 billion USD, presents Mythos’s capabilities to the public. The concerns raised by The Guardian have in fact received concrete responses from bankers and regulators. US Finance Minister Scott Bessent called bosses of major American banks, including Goldman Sachs and Citi, to discuss the cyber risks posed by Mythos. In the UK, regulators have included this issue in the agenda of the Cross Market Operational Resilience Group meeting, a high-level forum involving senior officials from the Treasury, Bank of England, Financial Conduct Authority, and National Cyber Security Centre.