A group of seven Republican senators has urged the US Department of Commerce to investigate the activities of the Chinese company DeepSeek, which has developed an open artificial intelligence model, R1. Lawmakers expressed concerns that this technology could be used to support military operations in China and poses a risk to the national security of the United States due to potential data transfer and the possibility of misuse.
This is reported by Business • Media
Concerns About Open Source and Military Collaboration
In their letter, the senators emphasized that DeepSeek already has a history of collaboration with military and intelligence agencies in China, and they believe this collaboration may continue. They pointed to the risks associated with the open-source nature of the AI model, particularly the potential for uncontrolled generation of harmful content and the deployment of the model without proper security testing.
“DeepSeek has a history of collaboration with military and intelligence agencies in China and is likely to continue it.”
Experts also note that open access to model weights could lead to remote code execution, data theft, or data tampering, and the lack of transparency regarding the origins of the AI could jeopardize confidential business information. One example of the negative use of R1 is the creation of instructions for social campaigns encouraging self-harm among teenagers and even the development of biological weapons.
Discussion on Restrictions and the Future of AI
Senators emphasize that issues of security and protection against information leaks must remain a priority for such technological products. At the same time, they describe the development of AI for business and consumers as a key area in technological competition with China. Back in April, the US Congress, during the Trump administration, considered the possibility of banning DeepSeek due to concerns about strengthening China’s position in the field of artificial intelligence.
Some experts agree on the necessity of restrictions in the public sector; however, they warn that strict bans could lead to the emergence of an oligopoly among American AI developers, slowing down innovation and complicating the decentralization of the industry outside the United States.
“Bans will lead to an oligopoly among American AI companies, which could slow innovation and create new problems,” added Anderson, emphasizing that this also complicates the decentralization of AI development outside the US.
Criticism regarding the security of DeepSeek has also come from prominent market representatives: Anthropic CEO Dario Amodei called DeepSeek the “worst” in data security tests among similar models.