State Department Confirms: Beijing Exploits U.S. Tech Platforms for Military Intelligence

| Published June 29, 2025

In a chilling confirmation of long-suspected digital infiltration, the U.S. State Department has sounded the alarm on a Chinese artificial intelligence company, DeepSeek, accusing it of covertly using American technology to power Beijing’s military ambitions. What looked like just another data-driven startup was, in fact, a well-embedded intelligence arm of the Chinese regime—quietly harvesting user data, bypassing export controls, and leveraging U.S. cloud services and microchips to support China’s People’s Liberation Army (PLA).

Revealed through procurement records, policy reviews, and growing bipartisan concern, this case exemplifies how adversarial powers are embedding themselves inside the very tech infrastructure the U.S. helped build. The revelation raises critical questions: Who exactly is behind DeepSeek? How did this operation go unnoticed for so long? And what does it say about America’s vulnerability in the AI arms race?

🕵️‍♂️ WHO – Key Actors Involved

1. DeepSeek

A Chinese artificial intelligence company, branded as a civilian enterprise but heavily entangled with China’s military apparatus.
DeepSeek has emerged as one of the largest Chinese users of U.S. cloud platforms, including services from Amazon Web Services (AWS), Microsoft Azure, and Google Cloud.
The company was repeatedly listed as a supplier for over 150 Chinese military research institutions, according to procurement records. This indicates its central role in PLA (People’s Liberation Army) tech modernization efforts.

2. People’s Liberation Army (PLA)

The armed forces of the Chinese Communist Party (CCP), and the ultimate beneficiary of DeepSeek’s activities.
The PLA allegedly used intelligence and data gathered through DeepSeek’s access to U.S. technology to enhance surveillance, simulate battlefield scenarios, and guide weapons development.
PLA entities purchased cloud-computing services and AI tools indirectly through DeepSeek, despite U.S. export restrictions on dual-use technologies.

3. U.S. State Department

A senior official from the State Department confirmed DeepSeek’s links to the PLA and issued a public warning about the use of U.S. tech platforms by hostile entities.
The official stated that the case reflects how China is “routinely evading export controls” by disguising military-linked entities as commercial operations.

4. U.S. Technology Providers

DeepSeek relied on major American companies for its computing backbone:
- Nvidia: Despite export bans, DeepSeek managed to obtain and operate powerful Nvidia H100 chips—one of the most advanced AI processors.
- Amazon, Microsoft, Google: Their cloud infrastructure was exploited to train DeepSeek’s AI models, making it easier for the company to scale intelligence work for the PLA.
- These companies are not accused of wrongdoing, but the case raises questions about loopholes and enforcement failures in current export and compliance systems.

5. Congressional Investigators and Analysts

U.S. lawmakers and intelligence oversight committees uncovered DeepSeek’s deceptive practices by reviewing procurement records and privacy policy disclosures.
Analysts flagged suspicious activity showing how Chinese firms harvest data from American users and funnel it back to Chinese telecom infrastructure, especially via China Mobile.

6. China Mobile

A state-owned Chinese telecom giant that served as an intermediary for data transmission.
Investigators found that user queries and behavioral data entered into DeepSeek’s AI platforms were being routed through China Mobile’s infrastructure and into Chinese government networks.

🧭 WHAT – What Happened?

The U.S. State Department confirmed that DeepSeek, a China-based artificial intelligence company, has been systematically exploiting U.S. technology platforms to aid China’s military intelligence efforts. This exploitation occurred in several overlapping ways:

1. Use of U.S. Cloud and AI Infrastructure

DeepSeek was found to be heavily reliant on American cloud platforms, including:

Amazon Web Services (AWS)
Microsoft Azure
Google Cloud

These platforms powered DeepSeek’s advanced AI models, which were used for data analysis, simulation, and likely surveillance purposes. Despite strict export regulations, DeepSeek gained access to powerful AI tools and infrastructure.

2. Access to Restricted Hardware – Nvidia Chips

Investigators discovered that DeepSeek obtained and deployed Nvidia H100 chips—high-performance processors used for training large AI models. These chips are officially banned from export to China, yet they ended up in DeepSeek’s servers. Their presence indicates the use of shell companies or third-party exporters to circumvent U.S. restrictions.

3. Data Harvesting at Scale

DeepSeek’s AI platform was available to global users, many of whom were likely unaware that their interactions—including searches, questions, and behavioral patterns—were being:

Collected without transparency
Analyzed for military-relevant insights
Routed through Chinese-controlled telecom infrastructure, specifically China Mobile

The platform essentially functioned as a global surveillance net, feeding the PLA with behavioral data useful for military simulations, targeting models, or psychological operations.

4. Repeated Engagements with the Chinese Military

Procurement records show that DeepSeek appeared more than 150 times as a supplier to PLA research institutes. This is not a coincidence or isolated incident—it’s part of a systematic collaboration between the company and military entities such as:

PLA’s National University of Defense Technology
Institutes working on AI-driven weapons systems, surveillance algorithms, and electronic warfare

5. Evading Export Controls

Despite being based in China, DeepSeek portrayed itself as a commercial AI company. This allowed it to:

Sign contracts with U.S. service providers
Lease computing power legally or via intermediaries
Appear “civilian” enough to escape enforcement red flags

The State Department confirmed that this deception represents a growing trend where Chinese firms “launder” military activity through tech start-ups, often using foreign shell companies in Southeast Asia and elsewhere.

🗓️ WHEN – When Did This Happen?

The exposure of DeepSeek’s covert operations unfolded over a period of several years, with a significant breakthrough and public confirmation coming in June 2025. Below is a breakdown of the timeline:

📌 Early Activity (2022–2023): The Quiet Build-Up

DeepSeek began scaling its AI capabilities using U.S.-based technology platforms, posing as a legitimate commercial tech company.
During this time, it secured contracts and computing access with cloud giants like Amazon, Microsoft, and Google—largely under the radar due to its seemingly civilian profile.
The company also began acquiring Nvidia chips, possibly through third-party or shell entities to bypass export restrictions that were already tightening under U.S. law.

🕵️ First Red Flags Emerge (2024)

In early to mid-2024, U.S. intelligence agencies and military cyber units—particularly those stationed in the Indo-Pacific, including Guam—detected suspicious cyber activity linked to Chinese state-backed actors.
DeepSeek’s name began appearing in procurement documents as a frequent supplier to PLA research institutes.
The U.S. Air Force and cybersecurity contractors began noticing unusually sophisticated AI-enhanced activities linked to China’s surveillance efforts, including behavioral modeling.

🧾 Procurement Records Uncovered (Late 2024)

Congressional investigators and analysts unearthed over 150 procurement records connecting DeepSeek to Chinese military and intelligence institutes.
Around this time, DeepSeek’s privacy policies and terms of use were scrutinized, revealing how user inputs could be lawfully (within China’s system) redirected for state use.

🚨 State Department Confirmation (June 2025)

In late June 2025, a senior U.S. State Department official publicly confirmed the long-suspected ties between DeepSeek and the People’s Liberation Army.
The official warned that DeepSeek’s use of U.S. cloud platforms and microchips posed a national security threat, and highlighted the company as an example of how China is routinely evading export controls through the guise of commercial operations.

This timeline shows how a seemingly legitimate tech firm was able to operate freely for years before the full extent of its military integration and exploitation of American platforms came to light. It also underscores how delayed enforcement and under-regulated access to sensitive technology created a window for China’s intelligence operations to flourish undetected.

📍 WHERE – Where Did This Happen?

The DeepSeek case spans multiple layers of geography: where the company is based, where it operates, where U.S. technology was exploited, and where the intelligence was ultimately routed. This is a transnational digital operation with both physical and virtual footprints across continents.

🇨🇳 China – The Core of the Operation

Headquarters: DeepSeek is based in China, operating under the guise of a commercial AI firm. Though no specific city is named in the report, most PLA-linked AI companies are centered in Beijing, Shenzhen, or Hangzhou—hubs for military-tech fusion.
End Destination of Data: All user-generated data—queries, prompts, interactions—collected via DeepSeek’s online platforms were funneled back to China. This was routed through China Mobile, a state-owned telecommunications giant.
Military Clients: The firm served over 150 PLA-affiliated research institutions, including military universities and labs working on advanced surveillance and combat technologies.

🇺🇸 United States – The Source of Technology and Data

Cloud Infrastructure Used: DeepSeek leveraged U.S.-based cloud service providers to run large-scale AI computations, including:
- Amazon Web Services (AWS) – HQ in Seattle, Washington
- Microsoft Azure – HQ in Redmond, Washington
- Google Cloud – HQ in Mountain View, California
Hardware Origin: DeepSeek obtained Nvidia H100 GPUs, which are manufactured and distributed by Nvidia Corporation, based in Santa Clara, California.
Data Source: Many of the platform’s users—possibly including unaware Americans—contributed behavioral data, search inputs, and interaction patterns, unknowingly feeding military intelligence pipelines.

🌏 Southeast Asia – The Shell Company Zone

According to intelligence findings, China has increasingly used shell companies based in Southeast Asian countries (possibly Singapore, Malaysia, or the Philippines) to:
- Mask military affiliations
- Purchase restricted components legally
- Circumvent U.S. export restrictions
These intermediaries allowed DeepSeek to access American technology while avoiding direct association with China’s military-industrial complex.

🇬🇺 Guam – Military Breach Site

Guam, a strategic U.S. territory in the Pacific, experienced attempted cyber intrusions in early 2024.
These intrusions were later linked to AI-enhanced cyber campaigns tied back to Chinese networks, where platforms like DeepSeek may have been used to simulate U.S. military communication patterns and infrastructure.

🌐 Online Platforms – The Global Interface

DeepSeek’s AI interface is accessible online and marketed globally, meaning the exploitation didn’t rely on physical presence.
Anyone using the platform—from students to researchers to businesses—could unknowingly have their input data processed, stored, and analyzed for military intelligence use in China.
This makes the operation both borderless and insidious, as users from North America, Europe, and Asia may have all contributed to data harvested by DeepSeek.

In short, DeepSeek’s operations were deeply embedded in U.S.-origin technologies, deployed globally through the internet, and routed back to China’s military command infrastructure—with shell entities helping cover the tracks in between.

❓ WHY – Why Did China (via DeepSeek) Exploit U.S. Technology?

At the core of this operation lies a calculated strategy by the Chinese government to leverage foreign—particularly American—technology to accelerate its military modernization, bypass costly R&D, and exploit the openness of global digital ecosystems. DeepSeek’s involvement wasn’t incidental; it was part of a deliberate and systemic campaign to bridge China’s technological gaps without provoking early detection or sanctions.

1. To Advance China’s Military Capabilities (PLA Modernization)

The People’s Liberation Army (PLA) is actively working to integrate artificial intelligence across all branches—cyber, air, land, sea, and space.
DeepSeek, under the cover of civilian AI research, served as a tool to develop:
- AI-powered surveillance systems
- Targeting algorithms for autonomous weapons
- Behavioral modeling for psychological and cyber operations
By using DeepSeek to collect behavioral and linguistic data globally, the PLA could train military AI systems using real-world data, especially from potential adversaries.

2. To Bypass U.S. Export Controls

Since at least 2018, the U.S. has imposed export restrictions on sensitive technologies—especially advanced AI chips like Nvidia’s A100 and H100 series.
However, China has developed tactics to evade these restrictions, such as:
- Using shell companies in Southeast Asia
- Misclassifying the end use of components
- Posing as commercial firms (like DeepSeek) to access services from American cloud providers
Exploiting these legal and enforcement loopholes allowed China to gain access to tools that would have otherwise been denied under U.S. law.

3. To Weaponize Open-Source and Commercial AI

The U.S. tech ecosystem is built on openness—open-source software, scalable cloud services, and global user access. China is weaponizing this openness.
By presenting DeepSeek as a global AI platform, China:
- Collected user input from around the world—especially the U.S. and allies
- Processed that data with high-performance computing from American cloud services
- Analyzed and stored it for use in training PLA intelligence and decision-making systems

4. To Reduce Domestic R&D Costs and Time

Developing a large-scale AI model from scratch is resource-intensive, requiring:
- Massive amounts of data
- Expensive computing power
- Skilled AI researchers
By exploiting U.S. platforms and tools, DeepSeek avoided these costs:
- It used American-built infrastructure
- It reused and repurposed data from global users
- It bypassed the need to build domestic alternatives, speeding up the PLA’s tech development by years

5. To Obscure Military Ties Through Civilian Fronts

China’s national strategy—“Military-Civil Fusion” (MCF)—encourages private companies to serve dual-use roles: appearing civilian while providing direct support to military objectives.
DeepSeek is a textbook example:
- Marketed as a consumer AI company
- Hidden contracts with military clients
- Operated legally in open markets while secretly feeding data to the PLA
This allows the Chinese state to shield its intelligence activities behind a veil of commercial legitimacy, making it harder for foreign governments to respond quickly.

6. To Undermine U.S. Technological and National Security

At a broader level, China seeks to reduce U.S. dominance in the global tech and defense landscape.
By infiltrating and exploiting American platforms, China:
- Harvests U.S. intellectual property indirectly
- Builds rival AI systems faster and cheaper
- Gains insights into U.S. user behavior, research patterns, and tech development trends
This not only narrows the technological gap between the U.S. and China but also creates strategic vulnerabilities within America’s own systems.

In short, DeepSeek’s exploitation of U.S. technology is not just a case of corporate misconduct—it’s a manifestation of a state-directed strategy to rewire global power through technology. The motive is clear: leapfrog America in AI and military readiness, without paying the full price or facing early detection.

🔍 HOW DID THEY FIND OUT? – Uncovering the DeepSeek-PLA Connection

The discovery that DeepSeek was acting as a covert arm of the Chinese military intelligence system was the result of multi-layered investigation across government agencies, congressional analysts, cybersecurity teams, and private sector researchers. It was not a single moment of exposure, but a gradual assembly of red flags that, when connected, revealed a pattern of military infiltration masked by civilian AI development.

1. Procurement Records Tied DeepSeek to the PLA

Investigators examined Chinese government procurement databases—some of which are publicly accessible or leaked through intelligence-sharing partners.
DeepSeek was found to have been listed as a supplier in more than 150 transactions with Chinese military and defense research institutions.
These included contracts with:
- PLA’s National University of Defense Technology
- Institutes focused on surveillance, cyberwarfare, and battlefield simulation
The sheer volume of these transactions raised immediate red flags about DeepSeek’s true nature.

2. Review of DeepSeek’s Privacy and Data Collection Policies

U.S. analysts examined the terms of service and privacy policies of DeepSeek’s public-facing AI platforms.
They discovered language that explicitly allowed the company to:
- Collect, analyze, and store user inputs
- Share that data with “authorized partners” under Chinese law (a phrase that often includes state agencies)
Legal experts pointed out that, under China’s 2017 National Intelligence Law, any tech company must comply with government data requests, making user data accessible to the PLA or intelligence services.

3. Detection of Illegally Acquired Nvidia Chips

DeepSeek was confirmed to be using Nvidia H100 AI chips, which are banned for export to China under U.S. export controls due to their dual-use potential.
This prompted an investigation into how the chips were acquired.
Evidence pointed to:
- Shell companies in Southeast Asia that acted as middlemen
- Purchases made under false civilian pretenses
- Shipping routes that deliberately obscured final delivery locations
The presence of these restricted chips in DeepSeek’s server infrastructure suggested intentional evasion of U.S. export laws.

4. Military Cyber Activity Traced to AI Platforms

In early 2024, U.S. military intelligence teams detected AI-assisted cyber intrusions targeting Guam and other Pacific installations.
These breaches used linguistic patterns and behavioral models consistent with large-scale language model training—suggesting the intrusions were powered by advanced AI.
Analysts traced digital footprints back to Chinese networks and eventually flagged DeepSeek as a key tool for PLA cyber units.

5. Congressional Investigations and Hearings

A growing concern within Congress about Chinese AI misuse led to oversight committees demanding deeper audits of companies like DeepSeek.
Staffers and researchers working for national security subcommittees compiled data from:
- Public business registries
- Chinese academic publication partnerships
- Export license filings
They noticed anomalies in cloud usage patterns—such as DeepSeek’s disproportionately large consumption of computing resources relative to its market visibility.

6. Cross-Agency Intelligence Sharing

Information was pooled from the State Department, Department of Commerce, U.S. Cyber Command, and private sector cybersecurity firms.
Patterns of military alignment, shell company use, and infrastructure abuse were compiled into a classified report, which ultimately led to the State Department’s public confirmation in June 2025.

7. Internal Whistleblower Accounts and Leaked Communications

Though not fully confirmed, reports suggest that internal communication leaks from a Chinese tech affiliate or an overseas contractor provided additional context about DeepSeek’s true mission.
This may have corroborated what procurement and infrastructure data already revealed: that DeepSeek was operating as a civilian façade for military intelligence operations.

In essence, DeepSeek was exposed not through a single smoking gun, but through a web of data points—procurement documents, tech audits, chip tracking, policy reviews, and intelligence cooperation—that collectively dismantled its cover story. The case now serves as a blueprint for identifying other dual-use tech firms embedded in open systems.

⚠️ Implications – Why This Matters

The exposure of DeepSeek as a PLA-linked company exploiting U.S. technology platforms carries profound implications—not only for American national security but also for the future of global technology governance, digital trust, and military deterrence. This is more than a case of unauthorized data harvesting; it’s a clear example of how authoritarian regimes can weaponize open innovation ecosystems.

1. National Security Threat to the United States

By secretly embedding military intelligence operations within a publicly available AI tool, China effectively gained a digital window into American user behavior, research interests, and communication patterns.
The use of U.S.-based cloud platforms and AI chips to power these operations means that America unknowingly became an enabler of PLA tech advancement.
This raises urgent concerns about:
- The infiltration of sensitive sectors
- Possible modeling of U.S. infrastructure or personnel
- The erosion of strategic tech superiority

2. Urgent Need to Tighten Export Controls

The case revealed that current U.S. export control measures—especially regarding semiconductors, cloud services, and AI platforms—are outdated, porous, and reactive.
Despite restrictions on high-end chips like the Nvidia H100, DeepSeek was able to acquire them using intermediaries and front companies.
This demonstrates a pressing need for:
- Real-time end-user monitoring
- Enhanced vetting of cloud service clients
- Stronger enforcement mechanisms to ensure that export bans are not just symbolic

3. Challenge to the Civil-Military Divide in Tech

DeepSeek exemplifies China’s Military-Civil Fusion (MCF) strategy, which deliberately blurs the line between civilian innovation and military application.
Western democracies, which traditionally separate commercial and defense sectors, now face the reality that any foreign-facing tech platform may be a dual-use Trojan horse.
This calls into question how “civilian” any Chinese tech company can truly be if they are compelled by law to cooperate with state intelligence demands.

4. Global Data Sovereignty and Digital Trust at Risk

Foreign users—including Americans—unknowingly shared inputs and behavioral data with an AI tool linked to China’s military, highlighting how trust in open internet platforms can be easily exploited.
This may trigger:
- Stricter digital sovereignty laws (e.g., EU-style data localization)
- More country-specific AI model regulations
- Chilling effects on global AI collaboration

5. Precedent for Future Digital Espionage

DeepSeek sets a dangerous precedent: if successful, it becomes a template for other state-backed companies to follow.
Other adversarial regimes may attempt similar operations, leveraging open AI tools as low-cost, high-reward intelligence collection machines.
The case signals that future espionage may not come from spies or hackers—but from apps, platforms, and chatbots.

6. Pressure on U.S. Tech Giants

While not directly accused of wrongdoing, American tech giants like Amazon, Microsoft, Google, and Nvidia now face pressure to:
- Strengthen client background checks
- Report suspicious usage patterns
- Work more closely with U.S. intelligence agencies
This introduces a delicate balance between business expansion and national security responsibilities.

7. Broader Strategic Competition Between U.S. and China

The DeepSeek case underscores the broader reality: we are in a full-spectrum competition, where AI, cloud computing, and user data are as critical as missile systems.
The U.S. must not only safeguard its infrastructure, but also actively defend the digital front lines—before commercial openness becomes a strategic vulnerability.

Overall Takeaway:

The DeepSeek exposé is more than a story about one AI company—it is a stark reminder that in today’s world, the battlefield has expanded beyond borders and into the cloud. Under the guise of innovation, China has leveraged U.S. technology, hardware, and infrastructure to silently accelerate its military ambitions. What seemed like routine cloud computing and AI development turned out to be a well-disguised intelligence operation feeding the Chinese Communist Party’s strategic arsenal.

This case underscores a crucial reality: open societies and open technologies are vulnerable if adversaries are willing to exploit freedom for authoritarian ends. It calls for an urgent reevaluation of how America handles tech exports, digital surveillance, and AI regulation—not only to protect intellectual property, but to defend national sovereignty.

The DeepSeek revelation should not be treated as an isolated breach, but as a blueprint of the next-generation threat landscape. The choice now lies in how swiftly—and how seriously—the U.S. and its allies respond.