U.S. Issues Global Alert Accusing Chinese AI Startup DeepSeek of Intellectual Property Theft

The United States government has formally accused Chinese companies, notably the artificial intelligence startup DeepSeek, of illicitly appropriating intellectual property from American AI research laboratories. This development was revealed through a diplomatic cable dated April 24, which was disseminated to U.S. diplomatic and consular posts worldwide. The cable instructs officials to raise concerns with foreign governments about what it describes as the “extraction and distillation of U.S. AI models” by adversarial entities.

In the context of AI, “distillation” refers to the process of training smaller, less resource-intensive models using outputs generated by larger, more sophisticated AI systems. This technique is often employed to reduce development costs but, according to the U.S. warning, may be used without authorization to replicate proprietary technology.

The U.S. State Department has not issued an immediate public comment regarding the cable or the allegations. However, OpenAI, the creator of ChatGPT, has previously alerted U.S. lawmakers that DeepSeek was attempting to duplicate leading American AI models for integration into its own AI systems.

China has categorically rejected these accusations. The Chinese Embassy in Washington labeled the claims as “groundless” and accused the United States of attempting to impede China’s technological advancement.

In a related development, DeepSeek recently launched a preview of its new AI model, V4, which is optimized for chips manufactured by the Chinese technology company Huawei. This release follows DeepSeek’s prior statements that its V3 model was trained exclusively on publicly available data collected through web crawling, and that it did not intentionally incorporate synthetic data generated by OpenAI.

Concerns over data privacy and security have led several Western governments, as well as some Asian countries, to restrict the use of DeepSeek’s AI products by government officials and institutions.

The diplomatic cable emphasizes that AI systems derived from unauthorized distillation of U.S. proprietary models may appear competitive on certain performance benchmarks and offer cost advantages. However, these systems purportedly do not match the full capabilities of the original models and may lack critical safeguards designed to ensure neutrality, reliability, and ethical use.

This warning reflects broader geopolitical tensions surrounding AI technology, intellectual property rights, and national security interests. It underscores the challenges governments face in balancing innovation, competition, and protection of proprietary technologies in the rapidly evolving AI landscape.