Back

FINMA Warns Anthropic’s Mythos AI Tool Could Create Systemic Risk in Switzerland

According to FINMA, banks’ rapid use of uncontrolled AI models could expose software systems to mass zero-day vulnerabilities and enable immediate AI-driven exploitation across the financial sector.

Summary

FINMA said Anthropic’s AI tool Mythos could pose systemic risk to Switzerland’s financial system and warned that banks’ rapid adoption of uncontrolled AI models may expose software systems to mass zero-day vulnerabilities. The regulator said such flaws could be exploited immediately through AI-driven attacks, framing advanced AI as an operational and cybersecurity threat that could spread across financial institutions because of their broad dependence on shared software systems.

Terms & Concepts
  • Systemic risk: The danger that a disruption at one or more critical points could spread across the broader financial system rather than remain isolated.
  • Zero-day vulnerabilities: Previously unknown software flaws that attackers can exploit before developers or users have time to issue or apply a fix.
  • Artificial intelligence model: A software system trained on data to perform tasks such as generating content, analyzing information, or identifying patterns.