China’s DeepSeek poses a "profound risk" to U.S. nationwide safety as a consequence of its sturdy ties to the Chinese language authorities, in line with a bipartisan report launched by the Home of Representatives’ Choose Committee on China. The report raises issues about DeepSeek's data-gathering strategies, data manipulation aligning with Chinese language Communist Social gathering (CCP) propaganda, and potential violations of U.S. export laws.
When DeepSeek unveiled its chatbot powered by the DeepSeek-R1 mannequin in January 2025, it despatched shockwaves by the AI trade. Not solely did the mannequin rival main Western AI techniques in efficiency, however its creators claimed it might function at a fraction of the fee and computational assets.
The declare that R1 value simply $6 million surprised specialists, who questioned how DeepSeek constructed such a sophisticated mannequin so cheaply. Since its launch, debates have continued over the determine, together with doubts over hidden prices and undisclosed assets.
Regardless of the skepticism, the rise of DeekSeek R1 underscores simply how far and quick China’s AI ambitions have progressed. Many had lengthy believed the West held a major lead within the AI race, however DeepSeek’s arrival has solid doubt on that assumption.
Shutterstock
As one U.S. AI Govt said within the report, “Some within the trade have claimed that the U.S. holds an 18-month AI lead, however that obfuscates reality- it’s nearer to 3 months.”
In keeping with the report, DeekSeek “funnels Individuals’ knowledge to the Folks’s Republic of China (PRC) by backend infrastructure linked to a U.S. government-designated Chinese language navy firm.” DeekSeek maintains shut ties to {hardware} platforms related to the Chinese language authorities, in addition to Zhejiang Lab, a state-backed analysis institute.
Whereas most of the high Chinese language tech companies have ties to the federal government, the report warns that DeepSeek crosses a purple line by combining a number of high-risk elements, together with siphoning U.S. person knowledge and secretly shaping content material to align with Chinese language Communist Social gathering directives.
The report states, "DeepSeek represents a profound risk to our nation’s safety. Though it presents itself as simply one other AI chatbot, providing customers a solution to generate textual content and reply questions, nearer inspection reveals that the app siphons knowledge again to the PRC, creates safety vulnerabilities for its customers, and depends on a mannequin that covertly censors and manipulates data pursuant to Chinese language legislation."
One other main concern highlighted within the report is the sturdy chance that DeepSeek unlawfully used mannequin distillation strategies to develop its AI system, successfully extracting key elements from U.S. AI fashions with out authorization.
The report states that an unnamed OpenAI government instructed the committee that an inner overview discovered DeepSeek staff had bypassed safeguards in OpenAI’s fashions. They extracted reasoning outputs, which can be utilized in a way referred to as “distillation” to hurry up the event of superior AI capabilities at a decrease value. Which means DeepSeek might have illegally benefited from U.S. expertise.
Moreover, the report states, "Equally troubling, the mannequin seems to have been constructed utilizing stolen U.S. expertise on the again of U.S. semiconductor chips which might be prohibited from sale to China with out an export license and when it was launched, PRC-affiliated social media accounts amplified and celebrated the mannequin, in line with Graphika analysis."
Supply: Shutterstock
The report claims that DeepSeek operates by backend techniques linked to China Cellular, an organization with established navy ties. Since Chinese language legislation mandates company knowledge sharing with the state, this connection poses critical nationwide safety issues. Lawmakers warn that DeepSeek might perform as a covert intelligence-gathering instrument for the Chinese language authorities.
In late 2024, the U.S. launched stricter export controls to limit China from buying superior AI chips. Nevertheless, these restrictions don’t seem to have had a serious influence on DeekSeek’s skill to amass high-performance {hardware}. In keeping with the report, DeekSeek makes use of “tens of hundreds” of restricted AI chips from American semiconductor big Nvidia.
The report has positioned Nvidia below scrutiny, questioning how its AI chips ended up in China. The chip producer has been accused of creating new chips designed particularly to keep away from the export management ban. Moreover, it alleges that Nvidia utilized middleman international locations to bypass the ban, with Singapore rising as a key hub for rerouting restricted AI {hardware} regardless of commerce laws.
Nvidia is dealing with a tumultuous week, with its inventory dropping practically 7% on Wednesday following information that the Trump administration would require a brand new license for all accelerated chip shipments to China. Nvidia CEO Jensen Huang made a shock go to to Beijing final week. He reportedly met with a number of Chinese language officers and Deepseek founder Liang Wenfeng. His journey has drawn elevated scrutiny over Nvidia’s function in China’s tech progress and whether or not the corporate is discovering new methods to navigate U.S. export restrictions.
“The emergence of DeepSeek is a warning to U.S. coverage makers that the PRC stays able to quickly innovating in at this time’s most superior applied sciences regardless of U.S. efforts to cease them,” the report cautions. “As AI continues to advance in functionality, U.S. departments and businesses should enhance their piecemeal method to forestall strategic surprises that will show destabilizing.”
The authors urge U.S. policymakers to strengthen export controls on superior {hardware}, together with semiconductor chips, and enhance funding for the Commerce Division’s Bureau of Business and Safety (BIS). It additionally recommends distant entry protections for U.S. AI {hardware} and whistleblower incentives to show export violations.
To counter China’s AI developments, the report additionally suggests empowering U.S. businesses to trace AI companies early, enhancing interagency coordination, and incorporating AI into nationwide safety planning. The geopolitical panorama is heating up, with the U.S. imposing elevated tariffs on Chinese language tech imports and pressuring allies to curb enterprise with companies linked to the Chinese language navy or surveillance state.
