U.S. artificial intelligence company Anthropic has decided to share cyber defense vulnerabilities identified by its latest AI model with major financial authorities and central banks worldwide.
The Financial Times reported, citing two sources familiar with the matter, that Anthropic plans to brief member countries of the Financial Stability Board (FSB) on the capabilities of its latest AI model, Claude Mythos Preview.
This briefing is reportedly in response to a request from Andrew Bailey, the Governor of the Bank of England and FSB Chair. The FSB is a global financial oversight body that includes finance officials, central bank representatives, and securities regulators from the Group of 20 (G20) nations.
In South Korea, the Bank of Korea and the Financial Services Commission are active members of the FSB, raising the possibility that the country will assess the performance of Mythos during this opportunity. Last week, Michael S. Cellioto, Anthropic's Global Policy Lead, visited South Korea to discuss issues related to Mythos with domestic authorities.
Many FSB member countries are concerned that Mythos and AI models from other U.S. tech firms may expose vulnerabilities in the cyber defenses of lending institutions, potentially creating new risks for the global financial system.
Last month, Anthropic reported that Mythos had identified thousands of high-risk vulnerabilities across all major operating systems and web browsers, warning that exploitation of these vulnerabilities could have severe implications for the economy, public safety, and national security.
Due to the risk of misuse, Mythos has primarily been disclosed only to select institutions within the United States. This has raised concerns among companies and regulators outside the U.S. about varying levels of cyber defense across regions, according to the Financial Times.
In response, the FSB is preparing a report outlining 'sound practices' necessary for the integration of AI into the financial system. The FSB plans to release this report next month and seek public feedback.
However, it remains uncertain whether international cooperation on AI threats will be effectively achieved amid geopolitical tensions. Earlier this month, the International Monetary Fund (IMF) urged policymakers to strengthen international collaboration to address the cybersecurity vulnerabilities revealed by the latest AI models.
The IMF warned that new AI models could elevate cyber risks to potential macroeconomic shocks, emphasizing that cyber threats do not respect borders and that emerging and developing countries with limited resources may be more exposed to attackers targeting vulnerable regions.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.