Open weights vs. closed: An AI civil war's afoot, and the stakes are existential

The article explores the deepening divide between open-weight and closed AI models. Open-weight models, such as Moonshot AI's Kimi K3, are shown to be nearly as fast as proprietary models like Anthropic's Fable 5 and OpenAI's GPT-5.6. US officials have accused Moonshot of stealing technology via distillation, but many in the industry view open-weight methods as legitimate. Microsoft released a policy statement supporting open weights, arguing they allow startups, universities, and public institutions to access AI affordably. Microsoft is joined by Amazon, Nvidia, Google, and nearly 200 Silicon Valley startups in urging the Trump administration not to restrict Chinese open-source models. Nvidia launched the Open Secure AI Alliance to address security concerns, noting that closed systems are not inherently safer. China's leader Xi Jinping advocated for global collaboration in AI. The distinction between open weights and open source is clarified: open weights lack full training data and source code. Anthropic and OpenAI oppose open models, with Anthropic CEO Dario Amodei expressing concerns about authoritarian governments building powerful AI. The article concludes that the conflict ultimately revolves around who benefits from openness and defines safety.