The Trump Administration just launched its own plan for global AI dominance and what could go wrong? (www.techradar.com)

The Trump White House announced the Genesis Mission executive order: a government-led effort, run out of the Department of Energy under OSTP oversight, to build an “American Science and Security Platform” — a centrally owned AI stack that will train scientific foundation models on federal datasets and create AI agents to automate research workflows. The platform will combine high-performance computing, modeling and analysis frameworks, computational tools, domain-specific foundation models, and experimental/production tooling to accelerate work in advanced manufacturing, biotechnology, critical materials, semiconductors, quantum information science, and nuclear fission/fusion. This is significant because it marks the first explicit push for a U.S. government–owned AI platform intended to rival private models (e.g., OpenAI, Google) and to counter China’s long-running national AI program. Technical promise aside, the plan raises immediate governance, privacy and dual-use concerns: training on vast federal datasets (much of it about citizens) could create opaque, government-controlled models; sensitive domains like nuclear and biotech amplify risks of misuse or accidental harm; and the EO stresses competitiveness more than public welfare or ethical safeguards. Partnerships with universities and industry are mentioned, but heavy in-house development could create a “black box” platform that demands clear transparency, oversight, and strict safety controls to avoid dangerous or militarized outcomes.