Executive Summary
History is at increasing risk of repeating itself on a larger, unprecedented level. The 1929 global economic crisis and (rising) potential modern financial instability driven by advanced AI share a common threat: the inability of existing infrastructure and governing institutions to withstand rapid, catastrophic change. There is growing concern regarding the risks to global financial stability stemming from the intersection of advanced artificial intelligence (AAI) – such as Anthropic’s Mythos – and cybercrime. The advanced AI capabilities present a double-edged sword; from offering proactive security to triggering a global domino effect causing instantaneous damage.
Key Points
- AI as a Double-Edged Sword: Financial firms are rapidly integrating AI to identify vulnerabilities in operating systems and web browsers. However, AI can simultaneously exploit weaknesses in major banks and financial infrastructures, capable of triggering an instantaneous, unprecedented, global crisis.
- Escalating Cyber Risk Exposes Systemic Vulnerabilities: Financial systems are at increasing risk due to their interconnectedness and the sensitive nature of the data they manage. A successful attack on one institution can trigger a domino effect across the entire global system. Emerging and developing economies are disproportionately at risk of attack.
- Regulatory and Governance Challenges: The expanding scope of AI capabilities raises questions about whether countries can collaborate and share information rather than act independently. AI cannot be confined within borders. Fragmented and inconsistent oversight, particularly among countries competing in the international information (AI) race, increases the difficulty of coordinating a large-scale response.
Analysis
The conversation has shifted. It is no longer just about defending against hackers; it’s about preparing for advanced AI autonomy and exploitation of flaws across the system, especially the global, deeply interconnected, financial system.However, financial institutions are not isolated in this potential attack. It is every interdependent institution from government to healthcare to airlines that would follow in the domino effect . It would bring the 1929 global crisis into the 21st century on another, unprecedented level.
The IMF’s article highlights the growing complexity of cyber capabilities as advanced AI is increasingly deployed in financial systems, using the example of Anthropic’s Mythos to illustrate both the promise and peril of advanced AI technologies . On one hand, AI is a key part of cyber defence. Anthropic’s Mythos, a state-of-the-art generative AI model, has demonstrated remarkable capabilities in preventing fraud, identifying systemic vulnerabilities, and detecting anomalies in financial operations . However, while AI cyber capabilities are not (yet) widely available , it is clear that these promises will hold only if there is a strong investment in governance and oversight practices.
On the other hand, while AI can strengthen defensive capabilities, the stability implications of these new advanced AI tools ultimately change the risk equation. The unequal distribution of these new advanced defensive tools as well as insufficient and fragmented international cooperation, will leave numerous institutions, particularly in emerging economies, vulnerable to attack. Models such as Anthropic’s Mythos can give cyber attackers the advantage as discovering and exploiting systemic vulnerabilities occurs faster than correction . AI driven cyberattacks therefore hold the potential to evolve from an isolated security breach into a macro-financial crisis that undermines the entire global system.
Policy Implications
- Dual Usage – “if you can’t beat them, join them”: These advanced AI tools can enable both attackers and cyber criminals, but they are also essential for cyber defense. When AI is used offensively, defensive systems must employ similar sophisticated tactics to remain effective.
- Resilience-Based Policy Framework: Cybersecurity must be treated as a core global stability issue, particularly in the financial sector. Since breaches are inevitable, policy should prioritize resilience to contain incidents, implement controls that prevent local attacks from escalating, and ensure swift recovery. While prevention remains important, focusing on resilience is increasingly critical as threats evolve.
- Governance Challenges: Advanced AI tools are now central to the global information race, and the Mythos case demonstrates the risks of limited international cooperation. AI cannot be contained within borders, so as its capabilities expand worldwide, inconsistent or private oversight threatens the stability of a globally interdependent system.