Concerns are rising in the financial sector about the potential for AI-driven cyberattacks, particularly stemming from a model known as Mythos, developed by Anthropic. This advanced AI has been kept from public release due to its ability to identify thousands of security vulnerabilities, including critical “zero-day” flaws that have persisted undetected for many years across various operating systems and browsers.
To address these threats, Anthropic has partnered with major companies, including Microsoft, Amazon Web Services, and Apple. The collaboration includes a commitment of $100 million in usage credits along with $4 million for open-source grants aimed at rectifying these vulnerabilities. Currently, over 40 organizations, including several U.S. banks, have been granted access to Mythos, although banks in Australia, the UK, and Europe have not yet received access.
Discussions at the recent International Monetary Fund Spring Meetings underscored the banking sector's exposure to these emerging risks, particularly due to its dependency on aging technology. Meanwhile, Anthropic is reviewing reports concerning unauthorized access to Mythos but has found no evidence suggesting malicious intent. For consumers in regions with strong banking safeguards, such as Australia, risks are currently managed through government insurance schemes for deposits.