European regulators were largely left out of Anthropic’s Mythos rollout as the company briefed only U.S. partners and officials

Of eight European national cyber agencies contacted by Politico, only Germany's BSI said it had entered conversations with Anthropic about Mythos, and it had not yet been able to test the model, contrasting sharply with the U.K. where the AI Security Institute confirmed it had already tested the model and taken action on its findings. Anthropic's 12 closest launch partners are all U.S.-headquartered companies, and the company said it has been in ongoing discussions with U.S. government officials, with no equivalent outreach confirmed to European regulators. The situation has exposed what EU experts describe as a lack of influence over American AI companies and raised questions about whether the bloc's AI Act, which requires providers to address cyber risks, gives Europe any meaningful oversight over models that are restricted rather than released to market.