Open Weights, Tightened Safety: The AI Policy Clash Redefining U.S. Tech Power

Silicon Valley’s AI leaders are releasing competing blueprints for superintelligence, debating whether safety comes from broader access to powerful models or from containment and pacing tools. Hassabis proposes a federally overseen Frontier AI Standards Body, Huang champions open weights to boost defense and innovation, Amodei calls for mandatory safety testing and tighter controls (without a full ban), Zuckerberg advocates broad access to personal AI, and Altman seeks a middle ground that preserves openness while enabling government‑managed pacing. Rogue-model incidents and a frontier Chinese model with open weights raise the stakes, while a forthcoming U.S. voluntary framework on AI reviews could determine which risks are prioritized and how access to the strongest systems is governed, shaping competition with China.