Bank of England to Brief Banks on Cybersecurity Risks From Anthropic's Claude Mythos
UK regulators including the Bank of England, the Financial Conduct Authority, the Treasury and the National Cyber Security Centre are urgently convening with major banks, insurers and exchanges to assess cybersecurity risks posed by Anthropic's Claude Mythos Preview model. The experimental system can identify hidden software vulnerabilities faster than human experts, and officials say it has already flagged thousands of major flaws across operating systems, browsers and widely used enterprise software. The model is being deployed under "Project Glasswing", a controlled initiative granting select organisations access for defensive cyber purposes. US regulators have raised parallel concerns, making this a coordinated transatlantic response.
Why it matters for Asia
Asia-Pacific's biggest financial centres - Singapore, Hong Kong, Tokyo and Sydney - take strong cues from the Bank of England on systemic cyber risk. If the BoE issues formal guidance restricting how frontier AI models interact with banking infrastructure, expect the MAS, HKMA and FSA to follow with their own frameworks within weeks. For enterprise security teams across the region, the message is clear: audit your exposure to AI-driven vulnerability scanning now, before regulators mandate it.
