Old Huang is on Twitter now, and the first post is a bombshell. Collaborating with multiple organizations, he wrote an article that does not mention KIMI even once, yet every line is about KIMI. It expresses a negative stance toward Claude's closed-off approach and is worth reading.

"Open Weights and America's AI Leadership"

In the 1980s, early open-source software pioneers challenged the prevailing belief of the time—that software could only advance under strict corporate control of the code. This movement fostered a transparent ecosystem, allowing developers worldwide to study, modify, and improve software. Today, the software from the open-source community underpins much of the internet’s infrastructure and forms the systems that are relied upon by the largest tech companies, the U.S. military, and federal agencies engaged in critical tasks such as research and cybersecurity. Open source not only reduces software costs; it creates a shared knowledge foundation upon which generations of American engineers and entrepreneurs build their institutional sovereignty.

The U.S. now faces a similar choice in artificial intelligence. America's AI leadership does not depend on a single cutting-edge model, but on whether the U.S. can establish a strong, open ecosystem that permeates various industries. This is crucial for nationwide innovation and prosperity. It requires expanding accessibility to AI, encouraging competition, developing a strong application layer, and giving Americans greater control over the technologies they rely on. Open weights models—AI models that anyone can download, review, modify, and run on their own infrastructure—are a vital part of this foundation, as they make advanced AI more accessible, customizable, and widely available.

Open weights expand access to the AI economy. Startups, established companies, universities, and public institutions can build on advanced models without needing to train from scratch or pay cutting-edge model prices for every task. Open weights enable every organization to match the right model to the right task at the right cost—reserving cutting-edge capabilities for truly front-line problems and running efficient specialized models for other scenarios. This discipline is key to maintaining economic sustainability as AI extends to billions of everyday tasks. The way for America to win the AI era is by allowing AI to permeate workflows in factories, hospitals, farms, classrooms, and corner stores.

Open weights also strengthen competition, and competition is essential for ensuring that the benefits of AI are widely shared rather than concentrated in a few hands. By allowing numerous organizations to build, modify, and deploy advanced models, the competition generated by open weights occurs not only among model developers but also across cloud, chips, applications, and services. This competition fuels innovation, lowers costs, and distributes the benefits of AI broadly across the economy.

Open weights also give customers greater control. When organizations invest in AI, they want to ensure they are not locked into a single vendor and do not lose long-accumulated knowledge and capabilities. Open weights models provide this assurance: organizations can control their own data, assess and modify models as needed, and deploy them wherever business needs arise. And when organizations create value with AI, open weights allow them to truly own that value through self-improving models, proprietary capabilities, and accumulated knowledge—these are the very things that drive American sovereignty and prosperity.

Admittedly, open weights carry real and unique risks. Once released, weights are beyond the control of the original developers, and modified versions can be hard to track or retract. But the appropriate response to these risks is not to ban open weights. In a world where cyber attackers are also using advanced AI, defenders need similarly capable models to detect, simulate, and respond to emerging threats. Open models expand defensive capabilities, increase transparency, and allow vulnerabilities to be identified and fixed by multiple teams.

In fact, openness may be one of the most important paths to AI security. Relying solely on closed models is not inherently secure: they can be breached, misused, or malfunction in ways that are not apparent externally. Concentrating advanced AI capabilities behind a few closed models amplifies this risk—creating single points of failure, weakening competition, and leaving critical technologies in the hands of a few suppliers. Open weights models enable a broad community of researchers and developers to audit model behavior, discover vulnerabilities, develop protections, and continuously improve. Just as open-source software has proven that transparency can be more secure than secrecy, AI security may similarly depend on enabling more people to test and reinforce the models that society relies on. It supports rigorous benchmarking and evaluation, red team exercises, and safeguards based on real, verified harm—rather than the assumption that "closed is safer."

A robust AI ecosystem is not guaranteed to emerge. Policymakers now have a crucial window of opportunity: to expand computing power supplies for startups and researchers; to invest in shared training assets (datasets, tools, evaluation frameworks); and to avoid imposing premature restrictions on open models—such restrictions would stifle competition or push innovation overseas—to maintain a diverse frontier. These measures should also consider how to leverage strong application layers to expand the sovereign use of AI throughout the economy.

In shaping this ecosystem, policymakers should be careful not to conflate legitimate model development techniques with infringement. Distillation—using the output of one model to help train or improve another model—is a widely used technique in model improvement, evaluation, and validation. It continues the rich tradition of learning from, adapting, and improving existing technologies, which has driven innovation since the rise of the open-source software movement. In contrast, the illicit extraction of value from closed models is indeed a cause for concern. However, these concerns should be addressed through targeted legal and commercial frameworks, rather than imposing one-size-fits-all restrictions on technologies that play a critical role in AI innovation.

The AI era can be a prosperous one. Making the right choices, open weights AI can expand opportunities, strengthen competition, maintain America’s technological leadership, mitigate risks, and ensure that the benefits of this extraordinary technology are widely shared throughout the economy. Such a future is worth building, and America should lead its construction.

Signatories (25 organizations): American Innovators Network · Andreessen Horowitz (a16z) · Arcee AI · Arena · Black Forest Labs · Box · CrowdStrike · Dell · Emergence Capital · Hugging Face · IBM · Linux Foundation · Mariana Minerals · Meta · Microsoft · Mistral · Mozilla · NVIDIA · Palantir · Perplexity · Reflection · Replit · ServiceNow · Telnyx · Y Combinator

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。