DeepSeek: A Deep Dive

House Science, Space, and Technology Subcommittee on Research and Technology

2025-04-08

Loading video...

Source: Congress.gov

Summary

The hearing addressed the critical implications of China's DeepSeek AI models on American technological leadership and national security. Witnesses discussed the rapid advancements in Chinese AI, the competitive landscape, and potential vulnerabilities, while also outlining strategies for the United States to maintain its innovative edge. The discussion highlighted the urgency of robust federal action, sustained investment, and coherent policy direction to secure America's future in artificial intelligence.

Themes

DeepSeek's Emergence and Threat

The emergence of DeepSeek marks a concerning milestone as the first non-American reasoning AI model directly influenced by the Chinese Communist Party (CCP)[ 00:20:52 ]

Citation requires full transcript

Sign in to continue

Sign in
. This development raises significant risks, including potential CCP access to sensitive user data and the propagation of CCP values through global AI applications[ 00:21:20 ]

Citation requires full transcript

Sign in to continue

Sign in
. DeepSeek's models reportedly rival top U.S. systems in performance while requiring fewer resources, showcasing China's rapid progress in AI. Several U.S. government agencies have already banned DeepSeek on federal devices due to serious data privacy concerns. It is suspected that DeepSeek developed its models using American semiconductor chips, possibly acquired before export controls or through evasion tactics. Additionally, DeepSeek suppresses information on politically sensitive topics like Tiananmen Square and the Uyghurs, reflecting the CCP's propaganda efforts[ 00:23:17 ]

Citation requires full transcript

Sign in to continue

Sign in
. While DeepSeek reported a low training cost of $5.6 million for its R1 model, experts suggest the total cost for its AI training infrastructure and experimental runs was likely much higher, possibly hundreds of millions or billions of dollars.

Maintaining American AI Leadership

Maintaining American leadership in AI is crucial for both economic competitiveness and national security[ 00:20:41 ]

Citation requires full transcript

Sign in to continue

Sign in
. The U.S. must ensure collaboration between the federal government and American industry to set global AI standards[ 00:22:36 ]

Citation requires full transcript

Sign in to continue

Sign in
. A "light touch" governance model that fosters innovation and avoids burdensome regulations is essential to prevent stifling U.S. developers compared to competitors like China[ 00:23:31 ]

Citation requires full transcript

Sign in to continue

Sign in
. The proposed "Create AI Act," which establishes the National Artificial Intelligence Research Resource (NAIRR), is considered vital for providing academic institutions and researchers with the necessary computational and data resources to compete[ 00:23:49 ]

Citation requires full transcript

Sign in to continue

Sign in
. Recommendations for strengthening U.S. leadership include reinforcing export controls enforcement, fostering an open research environment, incentivizing transparency across the AI lifecycle, and establishing robust data protection regimes. American firms lead significantly in private AI investment compared to China, indicating a strong foundation for innovation[ 01:14:56 ]

Citation requires full transcript

Sign in to continue

Sign in
. Furthermore, securing every layer of the AI technology stack, including hardware and software, and focusing on reliable AI is necessary to counter potential backdoors in models. Accelerating R&D through open-source model development and ensuring adequate electricity infrastructure are also key to maintaining a faster innovation engine. Attracting and retaining top global talent in STEM fields is highlighted as a critical American advantage[ 01:24:13 ]

Citation requires full transcript

Sign in to continue

Sign in
.

Challenges and Concerns for U.S. AI

The U.S. is facing a "Sputnik moment" with China's rapid advancements in AI, emphasizing the need for a swift and effective response[ 00:41:05 ]

Citation requires full transcript

Sign in to continue

Sign in
. Critics noted a lack of timely and effective implementation in U.S. export control policies, which has allowed China to stockpile chips and reduce the impact of these measures. The proliferation of diverse and often conflicting AI regulations at state and local levels could stifle innovation and create significant barriers for small businesses attempting to enter the AI market. Concerns were raised that the current administration's policies, such as funding cuts for federal research, the firing of scientists, and actions that deter foreign students, are actively undermining U.S. leadership in AI research and talent development. The discussion also acknowledged that concerns about data privacy and transparency extend beyond DeepSeek, as many U.S. AI companies also lack adequate clarity regarding their data handling practices. There is a recognized need for better, more standardized mechanisms for evaluating and benchmarking AI systems, as current methods are complex and can be manipulated, highlighting a crucial role for institutions like NIST.

Tone of the Meeting

The overall tone of the meeting was concerned and urgent, with speakers frequently referring to the situation as a "critical juncture" and a "Sputnik moment" for American AI leadership[ 00:20:43 ]

Citation requires full transcript

Sign in to continue

Sign in
. The discussion was largely competitive and strategic, framing AI development as a "race" or "Cold War" with China and emphasizing the need for robust strategic responses[ 00:26:37 ]

Citation requires full transcript

Sign in to continue

Sign in
. While the initial statements demonstrated a bipartisan consensus on the importance of AI leadership and the competitive threat from China, later exchanges revealed partisan disagreements regarding the effectiveness of current U.S. administration policies, particularly concerning research funding and talent retention[ 00:20:41 ]

Citation requires full transcript

Sign in to continue

Sign in
. Despite these tensions, the meeting maintained a forward-looking and action-oriented perspective, with witnesses offering concrete recommendations for legislative and policy interventions.

Participants

Transcript

Folks, just to update you, we're waiting on our last witness.  He is stuck in security.  He is on his way up, so hopefully we'll get underway just in a moment here.
The committee will come to order.  Without objection, the chair is authorized to declare recess at any time.  Welcome to today's hearing entitled Deep Seek, A Deaked Dive.  So we'll start out by opening with a five minute statement from myself.  I'd like to welcome everybody to our first research and technology subcommittee hearing.  I look forward to engaging with our distinguished panel of witnesses on what to me is a critically important topic.   It's clear that artificial intelligence will have a profound transformative effect on our country.  My experience leading the Bipartisan AI Task Force last year strengthened my belief that maintaining American leadership in AI development and deployment is not only an economic imperative, but also a national security requirement that affects every sector of our economy and our society.   As we examine the implication of DeepSeek's recent AI models, our nation is at a critical juncture in the global artificial intelligence landscape.  The introduction of DeepSeek represents a concerning milestone.  It's the first non-American reasoning AI model.  This capability, pioneered by American companies, is now being replicated by a company directly influenced by the Chinese Communist Party.  This development should raise concerns for all of us.   We must consider what's at risk.  Americans and people worldwide are increasingly sharing their private and personal data with AI systems.  The deployment of DeepSeek provides the CCP with a backdoor to this sensitive information.  This risk will only grow as we enter the era of agentic AI, where AI systems will actively book our travel, manage our finances, analyze our health records, and handle other sensitive personal affairs on our behalf.   We cannot allow DeepSeq and other CCP-controlled entities access to this information.  However, there's also a silver lining in this situation.  DeepSeq reportedly distilled their models from open AI systems, demonstrating that Chinese AI development remains reliant on our innovations.   And furthermore, despite the claim that DeepSeq R1 achieves similar results to American models at a lower cost, Google recently announced its open-weight Gemma 3 model, which reportedly achieves 98 percent of DeepSeq R1's performance for just 3 percent of the cost.
American ingenuity continues to lead the way, but we cannot take our continued leadership for granted.  Open-weight models underpin much of the AI and technology infrastructure worldwide, including here in the United States.   If we allow China to surpass us in open weight models, we risk ceding leadership in global AI infrastructure to the CCP.   It's crucial that we understand the capabilities of these models, the CCP goals they could propagate, and their potential vulnerabilities in order to encourage the adoption of American models over those developed in China.  This is precisely why the federal government and American industry must collaborate to ensure continued American leadership in the development of AI standards.  If the United States does not set these standards, then China will.   China's approach to AI development has also raised serious ethical and security concerns, especially relating to the prevention of harmful applications of AI.  For example, according to an evaluation by Anthropic, DeepSeek's model was found to be the least effective at blocking information about bioweapons amongst all the models that they tested.  While Chinese AI has so-called safeguards against providing information about Tiananmen Square and the Uyghurs, it lacks safeguards against actual malicious uses of AI.   We must ensure that Chinese AI, which operates under these flawed standards, does not come to dominate the global market.  The United States must take the lead in developing the most advanced AI systems while also fostering a light touch governance model that safeguards against malicious use while simultaneously encouraging innovation.  We cannot afford to stifle our innovators with burdensome regulations when competitors like China are racing ahead with fewer constraints.   Promoting innovation and AI development is the key to maintaining American leadership in this field.  To support this, I, along with some of the people on our committee here, have introduced the Create AI Act again in this Congress, which establishes the National Artificial Intelligence Research Resource.  This will provide researchers and developers across the computational and data resources they need to create competitive American AI systems that embody our values rather than those of the CCP.
I'd like to thank our witnesses for being here today.  I look forward to your testimony and to having a productive discussion on this critically important topic.  So I'll yield back the balance of my time.  I now recognize the ranking member, the representative from Michigan for her opening statement.
Thank you, Chairman Obernolte, someone I like to recognize as a friend, actually, and someone who I've gotten to know over my years in Congress, over his years in Congress, and particularly in his leadership role last term in the Congress running the Artificial Intelligence Task Force.  And now here he sits in a chair that I was once privileged to sit in as the chairman of the Research and Technology Subcommittee.   hosting and bringing together our first subcommittee hearing of the 119th session of Congress.  A deep dive into what we call Deep Seek, which is a brand of artificial intelligence that has been developed by the Chinese Communist Party.   CCP, and I'm explaining that because we have an incredible group of teenagers in the audience here who hail from the great state, arguably the best state in the union, Michigan.  We have reverends from Detroit, rabbis from West Bloomfield, and students who represent the incredible Black Jewish Student Alliance.   that the Anti-Defamation League of America has brought together.

Sign up for free to see the full transcript

Accounts help us prevent bots from abusing our site. Accounts are free and will allow you to access the full transcript.

Sign up