The head of the IMF has a new entry on her list of systemic risks: Anthropic's Mythos.
IMF Managing Director Kristalina Georgieva told officials in Brussels on Wednesday that advanced AI models like Mythos "can be used to destroy the financial system" if they fall into the wrong hands. The warning came during her presentation of the Fund's annual economic assessment of the eurozone, territory where frontier AI does not usually make an appearance. Naming a specific model, rather than gesturing at the technology broadly, was a notable departure from the vague hand-waving that multilateral institutions typically prefer.
When the IMF speaks, finance ministries and central banks tend to listen. By threading this warning into a formal eurozone assessment, Georgieva is effectively arguing that frontier AI deserves to sit in the same risk category as failing banks and shaky government debt. That framing, if it sticks, invites a different kind of regulatory response than the copyright disputes and bias audits that currently dominate AI policy conversations.
Georgieva did not specify who the "wrong hands" are, and the IMF has no direct enforcement authority. For now, this amounts to a very credentialed alarm bell with no accompanying sprinkler system.