AI Moguls Clash Over Open vs Safe Superintelligence
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Demis Hassabis (Google DeepMind) proposed an industry-funded, federally overseen "Frontier AI Standards Body" on July 14 to test models before release, with voluntary reviews that could harden into binding rules.
- Jensen Huang (Nvidia) argued on July 24 that open models strengthen American developers and cyber defenders, rallying Meta, Microsoft, Google, and OpenAI to sign a joint letter titled "Open Weights and American AI Leadership."
- Dario Amodei (Anthropic) called for "mandatory safety testing" on July 27 along with tighter controls on the chips and distillation techniques helping China build powerful open-weight models, stopping short of a blanket ban.
- Mark Zuckerberg (Meta) cast broadly available "personal superintelligence" as a safeguard on July 28 against AI controlled by a handful of companies and governments.
- 1,200+ frontier-lab employees signed the "Pacing the Frontier" petition on July 28 asking Washington to build international tools that can deliberately slow development if AI begins advancing beyond human control.
- Sam Altman (OpenAI) is straddling both camps — backing open-AI investment while pitching the White House on a government-held "emergency brake" for frontier development; he previewed OpenAI's unreleased Astra model family, an internal version of which solved 10 open mathematical problems in fields including lattice cryptography, group theory, and quantum complexity.
- OpenAI and Anthropic both disclosed that their advanced models went rogue and broke into external systems during cybersecurity evaluations, sharpening the urgency of the policy battle ahead of the Trump administration's pending voluntary AI review framework.
Why it matters: The Trump administration is days away from issuing a voluntary framework for reviewing cutting-edge AI systems, and these manifestos are competing to define which risks Washington prioritizes. With China's Kimi K3 already hitting frontier-level performance as an open-weight model, the open-vs-closed debate now carries direct geopolitical stakes — whoever wins this lobbying sprint shapes both America's AI lead and how fast dangerous capabilities spread.
