US Calls Bank Bosses, AI Anthropic Model Makes Regulators Nervous
JAKARTA - The United States government is taking cyber risks from artificial intelligence seriously. Anadolu Agency, citing a Bloomberg News report on Friday, April 10, reported that US Treasury Secretary Scott Bessent and Federal Reserve Chairman Jerome Powell held a surprise meeting with Wall Street executives.
The meeting took place Tuesday at the US Department of the Treasury, Washington. The goal is to ensure major banks understand the potential threat from Anthropic's latest Artificial Intelligence or AI model, Mythos, and other similar systems.
According to the report, officials want to see if the largest banks in the US are ready enough to face the new threat from AI models with higher capabilities. The concern is not small. Regulators assess that the new generation of AI can be used to launch more complicated and more difficult to read cyber attacks. If that happens, not only one bank will be disrupted, but it can spread to the financial system.
The banks called are not small banks. They are big players with a wide impact. If one of them has a problem, the effects can quickly spread to the global financial system. The names mentioned include Citigroup CEO Jane Fraser, Morgan Stanley CEO Ted Pick, Bank of America CEO Brian Moynihan, Wells Fargo CEO Charlie Scharf, and Goldman Sachs CEO David Solomon. JPMorgan CEO Jamie Dimon is said to be absent.
Anthropic describes Mythos as a much more powerful model. If asked by users, this model is said to be able to find and exploit security gaps in various major operating systems and web browsers. This is what makes regulator concerns stronger. Smarter technology can also make risks more difficult to control.
Access to Mythos itself is still restricted. Anthropic only opens it to a small number of major technology and financial companies, including Amazon, Apple, and JPMorgan Chase. The scheme is run through an initiative called Project Glasswing, which is aimed at securing critical systems before similar models are released more broadly.
Anthropic also said it had discussed with US officials before the model was released, including its cyber capabilities to attack and defend. At the same time, however, the company is also facing the Trump administration in court. The dispute arose after the Pentagon designated Anthropic as a supply chain risk. Earlier this week, a federal appeals court temporarily denied Anthropic's request to halt the designation.
For large banks, increasingly sophisticated AI models are now counted as a new source of risk. Because, disruptions to one system can quickly spread to the wider financial system.