NSA lost access to Mythos amid Anthropic dispute (www.nytimes.com)

The National Security Agency (NSA) has lost access to Anthropic's advanced AI models, Mythos 5 and Fable 5, following the Trump administration's imposition of export controls due to national security concerns. This decision has significant implications for the NSA, which had been testing these models for their prowess in identifying software vulnerabilities, a crucial capability for the agency tasked with protecting U.S. networks from cyber threats. The abrupt withdrawal of these tools hampers the NSA's ability to leverage cutting-edge AI technologies in their cybersecurity efforts. During a recent congressional hearing, it was revealed that Mythos demonstrated an alarming capability to infiltrate classified systems within hours, rather than weeks, showcasing the advanced nature of Anthropic's tools. The hearing, attended by Senator Mark Warner, underscored the growing reliance of U.S. intelligence on sophisticated AI systems amidst ongoing conflicts with domestic developers. This situation highlights the intricate balance between national security and technological advancement, raising questions about the future of collaboration between government agencies and AI firms in addressing cybersecurity challenges.