Overview
Anthropic plans to grant European banks access to its Mythos AI model in the near term, three people familiar with the matter said, as global lenders work to evaluate the technology after a group of large U.S. banks received initial access.
Regulatory and security concerns
Cybersecurity specialists view Mythos as presenting material challenges to banking firms and their legacy technology stacks. Those concerns were voiced by regulators and policymakers at last week’s International Monetary Fund spring meeting in Washington, prompting warnings about the risks the model could pose to the financial sector.
Phased rollout and timing
Anthropic has already provided Mythos access to partners inside its Project Glasswing initiative and to roughly 40 additional organizations that build or maintain critical software infrastructure. A number of U.S. banks have been granted access to the model, while the broader industry continues to seek entry points.
One person familiar with Anthropic’s plans said the company intends to expand access to banks in Europe and the United Kingdom, among other organizations. That person, speaking on condition of anonymity, said the process includes checks intended to ensure the rollout is handled securely.
Timing estimates differ slightly among sources. One individual said European banks could receive access within days, while another indicated the process might take days or weeks. A separate media report had previously indicated Anthropic would make Mythos available to UK financial institutions in the near future.
Anthropic did not immediately respond to requests for comment.
Who has access so far
JPMorgan Chase, part of the Glasswing initiative, is the only bank Anthropic has publicly acknowledged as having access. Bank of America has been part of Glasswing from its inception and, according to a source familiar with the matter, has been testing Mythos internally. Other U.S. banks have more recently said they have obtained access as well.
Calls for broader access
German central bank chief Joachim Nagel urged that all institutions be allowed access to Anthropic’s Mythos model to preserve a level playing field and to reduce the risk of misuse. Regulators are accelerating scrutiny of the cybersecurity risks associated with the new AI model as more institutions seek to evaluate its capabilities.
Location dateline: NEW YORK/PARIS, April 21