US security agency quietly tests Anthropic Mythos AI to hunt software vulnerabilities
The US National Security Agency (NSA) is testing Anthropic's Mythos AI to identify software vulnerabilities, including in Microsoft products. The evaluation is part of a broader effort to assess AI-driven cybersecurity tools, amid concerns about their potential risks and benefits.
The US National Security Agency (NSA) is actively testing Anthropic's Mythos AI system to identify vulnerabilities in widely used software, including Microsoft products. The evaluation is comparing Mythos's outputs with existing internal tools and research methods. Mythos rapidly scans complex codebases and flags possible weaknesses, potentially accelerating a process that traditionally requires extensive manual effort. Anthropic has limited access to Mythos, citing the risk that its capabilities could be weaponised by hackers. The company's CEO, Dario Amodei, has engaged with White House officials to explore safe use cases for the government. The NSA's testing is unfolding amid political tension, with the Pentagon labelling Anthropic's technology a potential supply chain risk.
This content was automatically generated and/or translated by AI. It may contain inaccuracies. Please refer to the original sources for verification.