European finance ministers are pressuring Anthropic to give local companies access to Mythos artificial intelligence models, lest they be exposed to digital attacks and fall behind their U.S. counterparts.
“Europe needs to respond,” Spanish Economy Minister Carlos Cuerpo told reporters on Monday ahead of a meeting of Eurogroup finance ministers in Brussels. He said the EU must determine "how we can defend ourselves and ensure that our businesses have access to these models to protect them from potential risks."
Finance ministers will discuss the issue on Monday amid growing concerns that the artificial intelligence model could trigger unprecedented digital attacks if it fell into the wrong hands. But so far, European companies have not gained access to Mythos, leaving them unable to assess the possible impact of the yet-to-be-released tool.
Cuerpo warned that Mythos and other new models may be able to "find vulnerabilities or backdoors in almost all of our institutions — not just in the financial sector and enterprises, but in every industry."
To this end, he advocated that Europe consider “regulatory and legislative tools such as the Artificial Intelligence Act” – the EU’s legal framework for the development of artificial intelligence.
Cuerpo reiterated this position at the meeting, according to people familiar with the matter. They said Eurogroup President Kyriakos Pierrakakis acknowledged the problem was serious enough that finance ministers would need to discuss it again at a future meeting.
Finance ministers are also demanding more reliable information, complaining that Europe can currently only assess risks based on hearsay, people familiar with the matter said.
Speaking at the Economic Committee of the European Parliament, Luis de Guindos, Vice President of the European Central Bank, emphasized the need to take precautions in advance to protect the European payment system from any digital vulnerabilities that Mythos may expose.
These technologies “should focus on identifying flaws in the operating system,” he said. But he added that if bad actors obtained these vulnerabilities, "it could cause a lot of problems."