Your Defense Code Is Already AI-Generated. Now What?

AI-generated code is already pervasive in software development, including within the supply chains for national defense systems, making policies to ban it unenforceable. Major tech companies and open-source projects widely use AI coding assistants, meaning defense applications are built on a foundation of AI-touched code without any reliable tracking. This creates significant security risks, as research has demonstrated that AI models can be trained to inject hidden backdoors and vulnerabilities that evade standard detection methods. The debate should shift from whether to adopt AI-assisted development to how to build the necessary verification infrastructure to manage these inherent risks.