US government wants American open-source AI to become the global default

The United States just made its AI ambitions about as subtle as a neon sign in Vegas. Which is fitting, because National Cyber Director Sean Cairncross chose the Black Hat conference in Las Vegas to announce that the US wants its open-source AI models to achieve what he called “preferential adoption by planet Earth.” What Cairncross actually said Speaking on August 4, Cairncross laid out a vision where American open-source AI becomes the global standard, not through mandates, but through sheer competitive superiority. The pitch frames open-source development as an innovation engine that particularly benefits startups and smaller firms that can’t afford to build proprietary models from scratch. The policy direction connects to a broader AI executive order signed in June 2026. That order prioritizes voluntary security testing and information sharing over heavy-handed regulation. The voluntary testing program specifically excludes open-weight models, which are the models where the architecture and trained parameters are publicly available but the training data and methodology might not be. That creates an awkward gap. The government is championing open-source AI while simultaneously c...