Anthropic’s Mythos found security flaws in classified U.S. systems within hours during testing, official says

Earlier this month, Trump directed the company to stop foreign nationals from using its latest AI models