Mythos model is an Anthropic AI system that sharpens risk signals and testing for government and finance, sparking global scrutiny.
Five Eyes agencies warn frontier AI could dramatically accelerate cyber threats in months, not years, while open questions linger about how to regulate and secure the technology.
Anthropic has released the Mythos model to a limited group of firms under Project Glasswing and has warned it can find thousands of software vulnerabilities faster than humans. Regulators and finance leaders in the US, UK, EU and Canada have convened urgent meetings, wargames and briefings to assess risks and coordinate defensive access and rules.
Carney has visited Ireland ahead of the G7 in France, emphasising a vision of dense, ad hoc coalitions between Canada, Ireland and Europe as the post-Cold War order shifts. He has celebrated roots in Aughagower while outlining a push to diversify trade and strengthen ties beyond the United States.
Anthropic’s Mythos model has identified vulnerabilities in sensitive U.S. systems during Project Glasswing tests. Officials say it could not exploit them within hours, while tensions with the government rise as policy tightens access to Mythos and related models.