Prime Highlights :
- Anthropic will brief global financial regulators on the cyber risks linked to its Mythos AI model.
- Regulators fear AI systems can expose banking software weaknesses faster than institutions can respond.
Key Facts :
- The Financial Stability Board includes major economies such as the US, UK, China, Japan, and Germany.
- Anthropic has limited Mythos access to around 40 organisations, including Amazon, Microsoft, and JPMorgan Chase.
Background :
Anthropic has agreed to brief global financial regulators on the capabilities of its cyber-focused AI model, Mythos, after concerns grew over the technology’s ability to uncover weaknesses in banking systems faster than institutions can respond. The discussions will involve members of the Financial Stability Board (FSB), which is chaired by Bank of England Governor Andrew Bailey.
Regulators and central banks across major economies are increasingly worried that advanced AI systems could create large-scale cybersecurity risks for the global financial sector. Officials believe Mythos can identify critical software flaws at a speed beyond the capacity of many organisations to repair them. The FSB is also preparing guidance on the safe use of AI in financial systems, with a consultation report expected next month.
Anthropic released Mythos earlier in May as a specialised cybersecurity model. The company acknowledged that the AI system discovered thousands of severe software vulnerabilities across major operating systems and web browsers. It also demonstrated the ability to generate exploit tools capable of attacking those weaknesses automatically.
Company officials warned that organisations could struggle to fix vulnerabilities quickly enough if attackers used AI models like Mythos at scale. Policymakers in the US and UK have already urged banks and financial institutions to strengthen cyber defences and speed up software patching efforts.
Anthropic has restricted Mythos access to around 40 organisations, including Amazon, Microsoft, and JPMorgan Chase. Meanwhile, OpenAI has also launched a similar cyber-focused AI model, increasing pressure on regulators worldwide.
Experts said AI-enabled cyber attacks are becoming faster and more sophisticated, raising concerns about financial stability and national security.