According to TechCrunch, the Trump administration is looking into placing restrictions on DeepSeek, a Chinese AI lab. Officials are considering banning Americans from using DeepSeek’s AI models and limiting the company’s access to Nvidia AI chips. These potential restrictions come from national security concerns about user data possibly being transferred to Chinese government servers.
DeepSeek is a Chinese AI startup that began in 2023. It makes open-source AI models called R1 and V3 that work similarly to OpenAI’s ChatGPT. The company is known for creating powerful AI systems at lower costs. Training their V3 model cost about $5.6 million (about ₹46.7 crore). DeepSeek uses a special method called “Mixture of Experts” to make AI training more efficient. The company wants to make advanced AI available to more people around the world.
The Trump administration’s AI team includes AI Czar David Sacks and advisor Sriram Krishnan. They have changed the government’s approach to AI by focusing more on innovation than safety. They canceled President Biden’s earlier AI rules and are now working on a new AI Action Plan to compete with China. These potential restrictions on DeepSeek show how the administration wants to protect U.S. data and keep America ahead in AI technology.
Nvidia’s AI chips are very important for companies like DeepSeek. Think of these chips like powerful engines that make AI work. Models such as the H100 and H200 provide the computing power needed to run complex AI systems. It’s like having a super-fast brain for the AI. Currently, DeepSeek cannot get the most advanced chips due to U.S. export controls. They have to use less powerful H800 chips or older A100 chips they saved up before restrictions. Without these advanced chips, DeepSeek may struggle to grow and improve their AI models.
OpenAI has accused DeepSeek of stealing their technology. They claim DeepSeek used a method called “distillation” – repeatedly asking ChatGPT questions and using the answers to build their own similar system. This would break OpenAI’s rules for using their service. Interestingly, DeepSeek’s AI model sometimes mistakenly identified itself as being made by OpenAI, which OpenAI points to as evidence. Microsoft detected suspicious activity from DeepSeek accounts in 2024 and OpenAI blocked this activity.
- National security experts have serious concerns about DeepSeek:
- China’s laws require companies like DeepSeek to share user data with the Chinese government
- Experts found code linking DeepSeek to China Mobile, a state-controlled telecom company
- This connection could allow user data to be sent directly to Chinese government servers
- The data at risk includes what users type, files they share, and their profile information
- This could lead to spying, watching users, and spreading AI-created propaganda
The restrictions being considered against DeepSeek vary in severity. Currently, 21 state attorneys general are asking Congress to ban DeepSeek on government devices. A proposed law called the No DeepSeek on Government Devices Act (HR 1121) would stop federal agencies from using DeepSeek. Some countries have already blocked DeepSeek on government systems. While these initial restrictions focus on government use, the Trump administration is also considering broader measures that could affect all American users and businesses. However, as of April 2025, there is no nationwide ban on DeepSeek.
This situation shows how AI technology is becoming a focus of international competition and security concerns. For everyday people in India and around the world, these restrictions highlight how the AI tools we use might raise questions about who can access our data. As countries try to protect their technological advantages, we may see more limits on which AI services people can use based on where those services come from.
