U.S. Lawmakers Demand Probe Into China’s DeepSeek AI Over National Security Fears

In a sharp escalation of U.S.-China tech tensions, seven Republican senators have called on the U.S. Commerce Department to launch an investigation into DeepSeek, the Chinese AI firm behind the increasingly popular open-source model R1, citing serious national security and data privacy concerns.

The letter—sent on August 1 and made public this week—warns that DeepSeek’s R1 model could leak sensitive U.S. user data and potentially aid Beijing’s surveillance or military objectives.

Lawmakers describe the situation as deeply troubling, especially given the model’s open accessibility and unclear safeguards.

Official assessments referenced in the letter suggest that DeepSeek has previously supported, and may continue to support, the Chinese government’s military and intelligence operations. The senators argue that allowing DeepSeek R1 to proliferate poses an unacceptable risk.

One of the key concerns lies in the model’s open-source architecture, which allows anyone to download and modify its model weights—the underlying parameters that guide its behavior.

This setup, they argue, opens the door for hackers to execute malicious crimes such as remote code execution, data theft, model tampering, and other forms of adversarial misuse.

Given these vulnerabilities, lawmakers insist that there is an urgent need for stricter oversight to ensure that foreign-developed AI models operating in the U.S. do not expose sensitive data or enable harmful exploitation.

The Trump administration had reportedly considered banning DeepSeek during its term due to the model’s lack of safety guardrails. Critics say these weak protections have allowed R1 to generate harmful content.

Among the more disturbing use cases cited: the model has allegedly provided instructions for launching social media campaigns promoting teen self-harm, as well as step-by-step guidance to construct a bioweapon—sparking outrage from policymakers and AI safety experts alike.

Regulation vs. Innovation

While lawmakers stress the need for strong guardrails, some experts warn of unintended consequences if regulation overreaches.

Chris Anderson, CEO of ByteNova AI, cautions that sweeping bans could centralize control in the hands of a few major U.S. tech firms, leading to an AI oligopoly and stifling open-source innovation.

Still, Anderson admits that in federal and critical infrastructure contexts, even a low-probability exploit could result in catastrophic damage.

“When provenance and auditability are unclear. Enterprises risk unknowingly exposing sensitive data or enabling adversarial misuse.”

With growing bipartisan concern in Washington over the influence of Chinese technology, the outcome of any Commerce Department inquiry into DeepSeek could have wide-reaching implications—not only for national security and the U.S.-China tech race, but also for the future of open-source AI development in an increasingly decentralized and globalized landscape.