OpenAI has alerted American lawmakers about the Chinese startup DeepSeek, which it claims is attempting to replicate its ChatGPT models and other advanced AI technologies for its own training purposes. This warning was conveyed in an internal memo reviewed by Reuters.
The memo details what OpenAI describes as “long-running attempts to piggyback on the capabilities developed by OpenAI and other leading American AI laboratories.” This strategy involves a technique known as distillation, whereby a more powerful model assesses the outputs of a weaker model, thereby transferring knowledge from one to the other.
According to OpenAI, employees of DeepSeek have allegedly devised methods to circumvent access restrictions to its models, utilizing measures such as hidden third-party routers and various techniques to obscure the origin of their requests. The company claims that software code was created to automatically extract responses from American AI models, which could then be used in the distillation process.
This memo was sent to the U.S. House of Representatives Special Committee on Strategic Competition between the United States and the Communist Party of China. Neither DeepSeek nor its parent company, High-Flyer, responded to Reuters‘ request for comment.
DeepSeek, based in Hangzhou, gained significant attention in the market early last year when it introduced AI models that were reported to be approaching the capabilities of leading American systems. This development has raised concerns in Washington regarding the potential acceleration of China’s advancements in AI, notwithstanding existing restrictions.
OpenAI also noted that many Chinese large language models frequently neglect safety standards during both training and deployment phases. Despite this, some in Silicon Valley have previously offered positive evaluations of DeepSeek’s models, such as DeepSeek-V3 and DeepSeek-R1, which are available to users globally.
In an ongoing effort to protect its intellectual property, OpenAI has stated that it routinely removes accounts that may engage in attempts to distill its models to create competing products. This highlights the broader tensions in the technology sector, where AI capabilities are becoming increasingly critical to national security and economic competitiveness.
The implications of these developments extend beyond corporate rivalry, as they reflect a larger geopolitical struggle over technological leadership. As nations race to dominate the AI landscape, the role of regulatory bodies and international cooperation will be crucial in addressing the challenges posed by unauthorized replication and the ethical use of AI technologies.
See also
TurboCell’s Modular Power System Targets AI’s Urgent Infrastructure Shortage by 2026
Mistral AI Commits €1.2 Billion to Build AI Data Center in Sweden by 2027
Trump’s AI Czar Warns 1,200 State Laws Risk U.S. Leadership in Global AI Race
Global Leaders Gather at India-AI Impact Summit 2026 to Shape AI’s Future, February 16-20
