Zhipu AI: Not Relying on U.S. Large-Model Technology , Undeterred by U.S. Entity List, Will Continue to Participate in Global AI Competition
Zhipu AI is now the first Chinese LLM company on the entity list
Being included in the U.S. export control entity list, Zhipu AI became the first Chinese LLM company sanctioned by the U.S.
The U.S. Department of Commerce stated that the reason for adding Zhipu AI to the Entity List is that it was engaged in advancing Chinese military via developing advanced A research.
“These entities advance the People’s Republic of China’s military modernization through the development and integration of advanced artificial intelligence research. This activity is contrary to the national security and foreign policy interests of the United States under Section 744.11 of the EAR.”
Furthermore, Zhipu AI was given a special “Footnote 4” designation. Under this designation, license requirements for it apply not only to items of U.S. origin, but also to foreign-produced items containing U.S. technology, software, or components under EAR jurisdiction, as set forth in Section 734.9(e)(2) of the EAR.
Zhipu AI will face more stringent licensing requirements when importing U.S. items. Whenever these listed companies act as a transaction party, the U.S. government mandates that any item subject to the EAR requires a license, together with a “presumption of denial” review policy. This means that, should Zhipu wish to procure any item controlled under the U.S. Export Administration Regulations (EAR), they must apply to the U.S. Department of Commerce for a license, but approval is highly unlikely.
On May 21, 2024, the second Artificial Intelligence Safety Summit took place in Seoul, South Korea. Together with 15 leading AI companies from North America, Asia, Europe, and the Middle East, Zhipu AI reached a consensus on safe AI development commitments and signed the Frontier AI Safety Commitments. It’s the first Chinese AI company that has publicly signed the commitments.
Zhipu AI issued a statement late at night saying that it possesses its own core large-model technology, so being added to the Entity List will not substantially affect its business.
A Chinese WeChat blogger commented that this indicates how U.S. concerns have been aroused by the competitiveness of Chinese companies in the AI field.
“The Americans seem to believe that they “absolutely must block” these companies.
Otherwise, they wouldn’t strategically pick a time and use such a drastic maneuver to crack down on a relatively young firm in the global AI arena.
This also underscores a stark reality: the urgency of having independent, controllable technology is greater than ever.
The road ahead will be neither short nor easy.
It appears that the United States believes it can keep sealing off and isolating Chinese high-tech enterprises by leveraging its own technological dominance.
Yet every attempt to surround and contain us only forces us to stand firm under pressure and pursue independent R&D.If Huawei can do it, then so can Zhipu.”
Zhipu: Statement on Being Listed by the U.S. Department of Commerce
Beijing Zhipu Huazhang Technology Co., Ltd. (“Zhipu”) has taken note that the U.S. Department of Commerce’s Bureau of Industry and Security (BIS) intends to add Zhipu and its subsidiaries to the Export Administration Regulations Entity List. We strongly oppose this decision, which lacks factual basis.
Zhipu evolved from technological achievements at Tsinghua University. As one of China’s earliest large-model enterprises, Zhipu has been developing the GLM pre-training architecture since 2020. We released the billion-scale bilingual pre-trained model GLM-130B in 2022 and made it open source; in 2023, we launched the 100-billion-parameter foundational dialogue model ChatGLM; and in 2024, we introduced the next-generation foundational large models GLM-4 and GLM-4-Plus, benchmarking against the world’s leading standards.
Zhipu remains firmly committed to technological investment. Centered on our independently developed GLM series, we have built the open platform bigmodel.cn under the “Model as a Service (MaaS)” concept. This platform continues to drive innovative AI transformation across countless industries. On the consumer side, our product “Zhipu Qingyan” has helped tens of millions of users in China access a homegrown AI assistant, leveraging AI to serve society.
Given that Zhipu possesses end-to-end core large-model technologies, being added to the Entity List will not materially affect our business. Zhipu is capable of, and will be more focused on, providing world-class large-model technologies, products, and services to our users and partners. At the same time, we will continue to participate in the global AI arena, adhering to the highest safety standards and the principles of fairness, transparency, and sustainability, so as to drive forward the development of AI technologies.
We sincerely appreciate the enduring support from all sectors of society.
Beijing Zhipu Huazhang Technology Co., Ltd.
January 15, 2025
智谱:关于被美国商务部列入实体清单的声明
北京智谱华章科技有限公司(以下简称“智谱”)关注到,美国商务部工业和安全局(BIS)拟将智谱及子公司增列至出口管制实体清单。这一决定缺乏事实依据,我们对此表示强烈反对。
智谱由清华大学成果转化而来。作为中国最早的大模型企业,自2020年开始投入研发GLM预训练架构,2022年发布中英双语千亿级超大规模预训练模型GLM-130B并开源,2023年推出千亿基座对话模型ChatGLM,2024年推出新一代基座大模型GLM-4和GLM-4-Plus,对标世界先进水平。
智谱在技术上坚定投入。基于自主原创的GLM系列模型,以Model as a Service(MaaS)为理念构建的开放平台 bigmodel.cn,正持续为千行百业带来人工智能创新变革。C端产品智谱清言,帮助数千万中国用户使用国产AI助手,用AI服务社会。
鉴于智谱掌握全链路大模型核心技术的事实,被列入实体清单不会对公司业务产生实质影响。智谱有能力也将更专注地为我们的用户和伙伴提供世界一流的大模型技术、产品和服务。同时公司将继续参与全球人工智能竞争,坚持最高安全标准和公平、透明、可持续原则,推动人工智能技术发展。
非常感谢社会各界长期以来的支持。
北京智谱华章科技有限公司
2025年1月15日
Looking forward to your take in the ai diffusion thing!