Artificial intelligence company Anthropic has agreed to brief key finance ministries and central banks worldwide on cybersecurity vulnerabilities in the global financial system discovered by its latest AI model. The arrangement follows a request from Andrew Bailey, Governor of the Bank of England, who also chairs the Financial Stability Board. Bailey sought for Anthropic to present the capabilities of its new Claude Mythos preview AI model to FSB members. The Financial Stability Board is an international regulatory body comprising finance ministry officials, central bankers, and securities regulators from G20 nations, including the United States, the United Kingdom, Canada, France, Germany, Japan, Saudi Arabia, and Australia. Many members of the board are increasingly concerned that Mythos and similar AI models developed by other U.S. tech firms could identify weaknesses in financial institutions' cyber defenses, posing potential risks to the global banking system. Anthropic stated last month that Mythos has already identified thousands of high-risk security vulnerabilities across major operating systems and web browsers. The company warned that exploitation of such flaws could have serious consequences for economic development, public safety, and national security. Due to concerns about the significant risks if the technology falls into the wrong hands, access to Mythos is currently restricted to a limited number of organizations, predominantly within the United States. This has raised concerns among businesses and regulators in other regions about a growing disparity in cybersecurity preparedness. Numerous organizations worldwide are applying for access to Mythos or seeking information about its functions. Anthropic has agreed to provide high-level briefings to some non-U.S. entities, including the European Commission. Currently, only about 40 organizations are authorized to use Mythos, including Amazon, Microsoft, and JPMorgan Chase, which employ the model to identify and fix system vulnerabilities. At the request of the White House, Anthropic has committed to not widely distributing the model. The Financial Stability Board is preparing a report outlining practical guidelines for the use of AI in the finance industry, with a draft planned for public consultation next month. Both the FSB and Anthropic declined to comment on their recent communications. Regulators globally are urging banks and financial institutions to conduct comprehensive reviews of their cybersecurity systems, promptly install patches, and address vulnerabilities exposed by new AI models. The UK Treasury and financial regulators have recently urged institutions in the City of London to take proactive measures to guard against faster and more destructive cyberattacks enabled by advanced AI, aiming to mitigate cybersecurity risks. However, amid current geopolitical tensions, some regulators believe it will be difficult to establish a globally coordinated response to the security threats posed by artificial intelligence. Earlier this month, the International Monetary Fund called on policymakers to enhance international cooperation to address the cybersecurity vulnerabilities exposed by the latest generation of AI models. The IMF warned that new AI models have elevated cybersecurity risks to a level that could trigger macro-financial instability. IMF staff wrote in a blog post that cybersecurity risks know no borders. Emerging economies and developing countries, often with limited resources and weaker cyber defenses, are more likely targets for cyber attackers and would suffer more severe impacts.
Comments