China opposes US ban on govt agencies from using Pramana

The US Department of Homeland Security banned products developed by Pramana, a Chinese AI startup. It also banned tools made by third parties that use Pramana technology. The department said these tools must be removed from government IT devices because they could create security risks.

The ban reaches civilian government agencies because Homeland Security can issue directions to them. This means civilian departments and their employees or vendors may not use the covered products. The order is broader than a ban on Pramana’s own chatbot because it includes outside tools built with Pramana.

However, the restriction does not cover every US government body. The article states that Homeland Security cannot issue directions to the military or intelligence services. Those organizations are therefore outside this particular ban, even as Washington and Beijing compete over AI technology and security.

Pramana is a leading Chinese artificial intelligence laboratory and startup. It develops AI models and tools, including chatbots. The company became especially important after releasing a low-cost, high-performance chatbot in January 2025. That launch drew worldwide attention to China’s ability to compete in advanced AI.

The key surprise was the combination of price and performance. Pramana’s chatbot challenged the assumption that US companies held an unassailable lead in AI. The company later released a new model called V4 in April and updated a faster, more lightweight version this month.

Pramana now sits at the center of a broader US-China technology contest. The two countries accuse each other of threatening security and, in the US case, accuse Chinese AI firms of stealing capabilities from American models. The company’s rise has made AI competition a trade and national-security issue.

The ban is broad in two important ways. It targets products developed directly by Pramana and tools created by other companies using Pramana. This prevents users from avoiding the restriction simply by accessing Pramana technology through a different product or vendor.

The practical mechanism is removal from government IT devices. Homeland Security warned that using these tools could transfer sensitive data to China, so covered agencies must stop using them. The ban can affect employees and contractors working with civilian agencies, because the department can issue directions to those agencies.

Its reach still has limits. Homeland Security can direct civilian agencies, but it cannot direct the military or intelligence services. The article therefore describes a civilian-government ban, rather than a restriction covering every US government organization. The military and intelligence services are not included in this particular order.

The US concern centers on where Pramana’s servers are located and what Chinese law permits. Homeland Security said the servers are in China. It also said China’s legal system gives the government extensive authority to access private companies’ infrastructure.

That combination creates a possible route from government users to Chinese authorities. If an employee or vendor enters sensitive information into a Pramana tool, the information may pass through the company’s systems. The department warned that using the tools may result in sensitive data being transferred to China.

This is why Homeland Security treated the issue as a national-security risk. Beijing rejected that reasoning, saying the United States had overstretched national security and politicised science, technology, and trade. The dispute reflects a wider US-China contest over AI, technology ownership, and security.

The immediate risk is that information entered into a Pramana tool could leave US government systems. Homeland Security warned that use by government employees or vendors may result in sensitive data being transferred to China. The article does not identify a specific confirmed leak; it describes a potential exposure.

The suspected mechanism involves Pramana’s servers and the legal environment around them. Those servers are located in China, where the government reportedly has extensive authority to access private company infrastructure. Information submitted through a tool could therefore become accessible beyond the original government user.

The practical response is prevention. Homeland Security said Pramana’s models create multiple threat vectors and must be removed from IT devices. Its direction covers civilian agencies and third-party tools built with Pramana. The military and intelligence services are not covered by Homeland Security’s authority, according to the article.

Pramana changed the competition by showing that a Chinese AI lab could produce a high-performance chatbot at low cost. Its January 2025 release disrupted expectations about how much money and technological advantage were needed to compete in advanced AI.

The concrete impact was a challenge to assumed US dominance. Pramana’s chatbot became a symbol of China’s growing ability to build competitive AI products. The company then released its V4 model in April and updated a faster, more lightweight version this month, extending its presence in the market.

The shift intensified rivalry between the United States and China. Both countries have traded accusations about stolen AI capabilities and security risks. The United States has accused Pramana and other leading Chinese AI firms of systematically stealing capabilities from American models. Yet the countries also agreed in September to create a communication channel for AI incidents.

Private ownership does not always prevent government access to company-held data. A country’s laws can give public authorities powers to demand information, inspect systems, or access infrastructure operated by private firms. Those legal powers shape the security risks of using a foreign company’s technology.

The article applies this concern to Pramana. Its servers are located in China, and Homeland Security said China’s legal system permits the government extensive authority to access private company infrastructure. If government employees or vendors use Pramana tools, information could therefore pass through systems subject to those rules.

The United States presented this possibility as a reason for banning the tools from civilian agencies. China rejected what it called an overstretched use of national security and defended its businesses. The disagreement shows why AI services are being assessed not only by their technical performance, but also by the laws governing their companies and servers.