Payward will use Anthropic's Claude Mythos 5 AI model to identify and fix software vulnerabilities, with plans to deploy it into the company's environment in the coming weeks. The project aims to help selected companies leverage cutting-edge AI models to protect critical software and infrastructure from cyber threats. Payward will also share vulnerabilities found in third-party open-source software with the relevant project maintainers. Previously, the U.S. government had made the Mythos 5 model available to U.S. organizations operating and defending critical infrastructure.