美国政府最近加强了对人工智慧的审查，限制了对两个 Anthropic 模型的存取，并推迟了一个 OpenAI 模型的发布。这项监管转变为全球企业领袖带来了重大不确定性，关于他们能否持续存取关键的 AI 基础设施。一项行政命令建立了一个自愿性框架，要求 AI 公司在公开发布前，向政府提供长达 30 天的前沿模型存取权限，旨在监管这些快速发展的能力。
为了应对这种不确定性，Ion Stoica 和 Gautier Cloix 等行业专家建议企业透过利用多个 AI 模型（包括开放权重和封闭系统）来实现技术依赖的多样化。虽然 OpenAI 和 Anthropic 等开发商的封闭模型提供的透明度较低，但开放权重替代方案允许开发人员自订数值和权重。诸如 H Company 等公司目前正在微调这些开放模型，以维持技术自主性并在危机期间平衡风险，将多样化的模型格局视为拥有车辆的多种能源选择。
全球 AI 发展的格局竞争日益激烈，许多顶尖的开放 AI 模型源自中国，促使 Xi Jinping 等领导人强调在技术监管方面的国际合作。与此同时，诸如 Yann LeCun 等 AI 先驱透过 Project Tapestry 等倡议，提倡一种完全不同的训练范式。该项目旨在汇集来自国家、大学和行业的训练资料，以建立开放的基础模型，同时允许贡献者保留对其资料的控制权，为 AI 的未来提供了一个免费且不可或缺的替代方案。
The United States government has recently intensified its scrutiny of artificial intelligence by restricting access to two of Anthropic's models and delaying the release of one OpenAI model. This regulatory shift has introduced significant uncertainty for global business leaders regarding their continuous access to crucial AI infrastructure. An executive order established a voluntary framework requiring AI companies to provide the government with access to frontier models for up to 30 days prior to their public release, aiming to regulate these rapidly evolving capabilities.
In response to this uncertainty, industry experts like Ion Stoica and Gautier Cloix advise businesses to diversify their technological dependencies by utilizing multiple AI models, including both open-weight and closed systems. While closed models from developers like OpenAI and Anthropic offer less transparency, open-weight alternatives allow developers to customize numerical values and weights. Companies such as H Company are currently fine-tuning these open models to maintain technological autonomy and balance risks during crises, treating the diverse model landscape like having multiple energy options for a vehicle.
The global landscape of AI development is increasingly competitive, with many top open AI models originating from China, prompting leaders like Xi Jinping to emphasize international collaboration in technology regulation. Meanwhile, AI pioneers such as Yann LeCun advocate for a completely different training paradigm through initiatives like Project Tapestry. This project aims to aggregate training data from countries, universities, and industries to build open foundation models while allowing contributors to retain control over their data, presenting a free and indispensable alternative for the future of AI.