The U.S. Department of Homeland Security issued a directive ordering all federal civilian agencies to remove artificial intelligence technology developed by Chinese startup DeepSeek from IT systems and ban future contracts or third-party integrations. Citing cybersecurity risks and sensitive data transfer to China, the ban affects agencies like State and Justice while excluding Defense. DeepSeek's low-cost models—trained for $6 million compared to OpenAI's $100 million GPT-4—have seen rapid adoption among U.S. startups amid IP theft allegations.
US federal agency DeepSeek ban
- ▪A Department of Homeland Security directive issued Monday prohibits federal civilian agencies from procuring future products, services, or third-party offerings that integrate Chinese startup DeepSeek's artificial intelligence models
- ▪An October 19, 2026 U.S. Department of Homeland Security notice ordered federal civilian agencies to remove DeepSeek technology from information systems over cybersecurity concerns that models could transfer sensitive data to the People's Republic of China
- ▪The Department of Homeland Security's ban on DeepSeek AI, announced Monday, took effect at the end of July 2026 and applies to agencies such as Treasury, State, and Justice, while excluding Defense Department agencies
US concerns over Chinese AI
- ▪U.S. President Donald Trump has positioned China as the primary U.S. rival in artificial intelligence development, framing AI leadership as crucial to the nation's future
- ▪U.S. artificial intelligence companies Anthropic and OpenAI claimed that Chinese AI firms, including DeepSeek, essentially copied their proprietary AI models
High user demand for DeepSeek
- ▪High demand for Chinese artificial intelligence startup DeepSeek's models has caused company servers to crash at times since its initial public release in 2025
- ▪Chinese artificial intelligence startup DeepSeek's free chatbot application ranks among the most downloaded apps in the Apple App Store
DeepSeek AI model cost efficiency
- ▪Chinese AI startup DeepSeek trained the debut version of its artificial intelligence model for approximately $6 million, compared to roughly $100 million spent by OpenAI to train GPT-4
- ▪Using DeepSeek's open-source artificial intelligence models costs developers a fraction of the price charged for closed models such as Google's Gemini, Anthropic's Claude, or OpenAI's Codex
Story comments
Loading comments…