EMET NEWS PRESS

How Does China Access Research?

Thesis: China uses international research as a capability source. Civil-military fusion links universities, state institutes, industry, intelligence services and the People’s Liberation Army.

Substance: China’s access occurs through legal and semi-legal transfer forms. Doctoral programmes, visiting-researcher stays, joint publications, conferences, code repositories, research data, laboratory access and follow-on careers generate methodological capability gain. Classic leakage of classified material is not required.

The People’s Liberation Army requires cyber capabilities for reconnaissance, C2 protection, access to adversary systems, information operations, electronic effects and resilient combat operations. Research fields such as software quality, app behaviour, AI security, attack detection and cryptography sit directly inside this capability profile.

Formal partner labels carry limited security value. A civilian institute in China can enable military follow-on use through state programmes, personnel pathways, third-party funding, patents, procurement chains or publication cooperation. Legal form is not decisive; transfer value is.

Net assessment: China does not require covert full access if open research structures deliver equivalent capability gain. Germany’s risk lies in legal transfer with military follow-on use.

Which Transfer Pathways Are Operationally Relevant?

Thesis: Core transfer runs through people, methods, code and networks. Theft of documents is secondary.

Substance: Personnel transfer provides procedures, working methods, research routines, laboratory practice, network access and tacit methodological knowledge. This knowledge remains usable after return to Chinese institutions.

Publication transfer delivers reproducible methods, evaluations, datasets, attack models and defensive architectures. Cybersecurity research often publishes sufficient technical depth to enable replication and further development.

Code and tool transfer enables direct reuse. Open-source code from security and AI research can be globally adapted, combined and operationalised. A classical export event does not occur.

Cooperation networks generate trust, recommendations, supervision relationships, access to non-public contextual knowledge and durable expert links. For intelligence services, military-adjacent research actors and defence industry, such networks have higher value than one-off datasets.

The arrests in May 2026 confirm this mechanism. According to the Federal Prosecutor General, two German nationals allegedly built contacts to German universities and research institutions for a Chinese intelligence service. The target was information on advanced technology with potential military value, including aerospace, computer science and AI. Several scientists were allegedly invited to China under civilian cover. The operational core was contact-building, expert access, invitation, network formation and technology assessment.

Net assessment: Transfer is cumulative. Person, project, publication, code, invitation, network and return pathway together generate capability gain. The May 2026 case is a German case anchor for precisely this access method.

What Capability Gap Does the CISPA Complex Expose?

Thesis: CISPA exposes a control gap in Germany’s cyber research space. High-grade cyber capabilities are being scaled with state funding; international cooperation risks are not systematically assessed for military follow-on utility.

Substance: CISPA is a Helmholtz Centre for Information Security. Its research fields include cybersecurity, trustworthy AI, software analysis, privacy, cryptography and secure system architecture. In 2019, CISPA cooperated with Chinese research institutions, including the Institute of Software of the Chinese Academy of Sciences and the Institute of Intelligent Software in Guangzhou. The cooperation covered researcher exchange, joint projects, conferences and seminars.

The funding dimension is strategically relevant. From 2029/30, CISPA is set to receive around EUR 45 million in additional annual funding. Saarland is providing up to EUR 350 million for the new campus in St. Ingbert. This creates a national cyber centre with high personnel, method and infrastructure value.

Open sources do not conclusively verify that German public funding directly financed Chinese military programmes. The risk finding remains robust: cooperation in security-relevant cyber research can feed into Chinese capability chains because China structurally links civilian and military exploitation. For the named Chinese partners, institutional affiliation is openly verifiable; a concrete PLA subordination of these partners is not verifiably established in open sources. That is the point of the security requirement: review must assess military follow-on utility, not only formal partner structure.

Net assessment: CISPA represents a German Dual-Use control deficit. The core issue is not one isolated project, but the absence of deep review of partners, personnel, methods, code and end use.

Why Is Cybersecurity Militarily Exploitable?

Thesis: Cybersecurity is a core Dual-Use field. Defensive research improves protection; the same methods improve attack preparation, target analysis, exploit development and military network resilience.

Substance: Software analysis identifies vulnerabilities, attack surfaces, unwanted behaviour and critical code paths. This protects civilian systems, but also supports offensive target reconnaissance and exploit development.

Cryptography and secure system architecture protect communications, but also increase the endurance of military command networks. AI-security research improves model robustness and abuse detection, but also stabilises offensive AI tools, target classification, data fusion, malware adaptation and disinformation instruments.

Cyber operations against military systems and civilian infrastructure sit inside the same technical space. Air defence, energy supply, railway control, satellite communications, logistics software, cloud infrastructure and defence production depend on software-enabled resilience.

Net assessment: Cybersecurity research has direct military follow-on utility. Open publication and international cooperation can generate capability gain without the leakage of classified material.

Which German Control Gaps Remain?

Thesis: Germany’s control architecture covers classical export goods more effectively than research, code and personnel transfer in cyber and AI fields.

Substance: BAFA export control covers goods, software, technology, technical assistance, publications and visiting researchers where control triggers apply. The German China Strategy identifies risks from illegitimate influence, one-sided knowledge and technology transfer and civil-military fusion. The domestic intelligence service warns against Chinese espionage targeting business, science and high technology. The EU, DAAD, HRK, DFG/Leopoldina and BMBF have developed guidelines, recommendations and awareness frameworks.

The enforcement gap remains. Publications, open software, conference papers, academic visits, joint basic research and follow-on careers often sit below licensing thresholds. In cyber fields, relevant transfer occurs exactly there. A mandatory situation picture covering partner profiles, personnel pathways, code releases, datasets, end use and return structures is missing.

The architecture remains distributed: university autonomy, academic freedom, export control, domestic-intelligence awareness, ministerial funding and institutional compliance operate in parallel. No unified high-risk review path decides before project launch on military follow-on utility.

Net assessment: Germany has building blocks, but no closed security regime. The existing architecture warns, advises and controls selectively; it does not reliably block risky transfer pathways.

What Role Do Federal and State Funds Play?

Thesis: Public funding increases target value and protection duty. State-funded cyber excellence must be secured as a national capability base.

Substance: CISPA receives public baseline and expansion funding. From 2029/30, around EUR 45 million in additional annual funding is expected. The St. Ingbert campus is being built with up to EUR 350 million in state funding. These funds create personnel, infrastructure, laboratory environment, international visibility and cooperation attractiveness.

The larger the state-funded cyber centre, the higher its access value for foreign intelligence services and military-adjacent research actors. Protection measures must scale accordingly: partner review, risk classification, reporting duties, security advice, project monitoring, code and data review, exit criteria and enforceable sanctions.

Funding can trigger security obligations. In high-risk fields, this is not a restriction on research; it is protection of national cyber capability.

Net assessment: Cyber funding without transfer control increases the value of the target. Funding decisions must trigger security architecture.

What Military Effects Can China Gain?

Thesis: Chinese capability gain lies in cyber operations, C2 resilience, AI-enabled target processing, information operations and protection of its own military networks.

Substance: Software and vulnerability research accelerates attack preparation. Automated analysis assesses target systems faster. Cryptography and secure architecture increase the robustness of Chinese command and communication networks. AI-security methods stabilise offensive AI tools, model protection, target classification and data fusion. Research on app behaviour supports mobile target analysis, malware detection and platform assessment.

Germany faces counter-risk across critical infrastructure, defence industry, government networks, military logistics and communications systems. Cyber capabilities support hybrid operations, sabotage preparation, disinformation, economic espionage and crisis escalation below the threshold of war.

Realisability is high. Open research, publication, code and training are sufficient for methodological capability gain.

Net assessment: The risk is operationally real. It affects national security capability, military resilience and protection of critical infrastructure.

Which Countermeasures Are Required?

Thesis: Germany needs a binding high-risk review path for security-relevant research. Advisory structures are insufficient.

Substance: Existing instruments do not create closed enforcement. BAFA provides export-control frameworks and academia guidance. HRK and DAAD provide China-related guiding questions. DFG and Leopoldina address security-relevant research through risk review and self-regulation. The EU research-security recommendation calls for risk management. BSI and domestic intelligence raise awareness of cyber and espionage risks.

For cybersecurity, AI, cryptography, quantum, space, robotics, semiconductors and security-relevant software, every high-risk cooperation requires a risk dossier before project launch. Mandatory contents: partner profile, PLA/MSS proximity, third-party funding, military publication patterns, patents, personnel access, code releases, data releases, return pathways, end-use risk and termination criteria.

The review architecture must connect research funding, BAFA, BSI, domestic intelligence, BMBF, BMWK and scientific organisations. Funding must be tied to security conditions. Open-source and publication decisions in high-risk fields must be assessed for military exploitability.

Net assessment: Germany needs enforcement, not more awareness. The standard is prior review, ongoing monitoring, documented termination criteria and sanctionable funding logic.

Situation Assessment

Germany is building CISPA into a cyber capability of strategic rank. The transfer pathways of this capability are not protected by a unified security regime. Partner review, personnel transfer, code release and end use are not systematically controlled for military follow-on utility.

China uses high technology as a systematically exploitable military resource. Civilian research institutions, state institutes, industry and military modernisation are tightly linked. Formal partner labels carry limited security value.

CISPA remains a central German cyber actor. The security finding concerns risk governance, not research excellence. Target value increases with funding volume, personnel strength, international visibility and methodological depth.

Federal government, states and research organisations must treat cyber research as a national capability base. Funding, internationalisation, export control, research security and intelligence assessment belong in the same control framework.

Final Assessment

The situation holds if Germany subjects high-risk cooperation in cybersecurity, AI and security-relevant software to binding review. Required measures are partner screening, end-use analysis, personnel review, code and data control, reporting duties, termination criteria and a joint review architecture involving research funding, BAFA, BSI, domestic intelligence, BMBF, BMWK and scientific organisations.

The situation deteriorates if Germany continues to finance cyber excellence while assessing international follow-on exploitation only formally. The research space then remains usable for China through personnel, methods, code, publications and networks.

Net assessment: The CISPA complex exposes a strategic weakness in German security and research policy. Cyber funding is capability development. Capability development without transfer control can strengthen adversarial cyber power.

Glossary

Civil-Military Fusion
Chinese strategy linking civilian research, industry and military modernisation. It complicates separation between civilian cooperation and military follow-on use.

Dual Use
Technology or research with civilian and military utility. In cybersecurity, AI, cryptography and software analysis, the distinction is structurally weak.

End-Use Review
Assessment of who ultimately uses a technology, method, software or research output and in which capability context it can be applied.

C2 Resilience
Resilience of command-and-control and communication systems against disruption, failure, cyberattack and electronic effects.

Transfer Pathway
Route through which knowledge, methods, code, data, personnel or networks move from one research environment into another capability structure.

High-Risk Cooperation
Research cooperation in a security-relevant technology field involving partners, personnel pathways or end-use risks that may enable military or intelligence follow-on exploitation.

References

CISPA Helmholtz Center for Information Security
A new cooperation with top research institutes in China; information on the Institute of Software, Chinese Academy of Sciences, and the Institute of Intelligent Software Guangzhou, 30 April 2019.
cispa.de/en/cooperation-research-institute-china

CISPA Helmholtz Center for Information Security
CISPA to receive around EUR 45 million in additional annual funding from 2029/30, 2 June 2026.
cispa.de/en/funding-increase

Institute of Software, Chinese Academy of Sciences
Institutional profile and research fields, including Internet Software Technologies, Intelligent Software Research Center and Data Science.
english.is.cas.cn

Institute of Software Application Technology, Guangzhou
Institutional profile on applied software technology, smart-city research and innovation teams.
gzis.ac.cn/en

Federal Prosecutor General at the Federal Court of Justice
Two arrests on suspicion of intelligence-agent activity for a Chinese intelligence service, 20 May 2026. www.generalbundesanwalt.de/SharedDocs/Pressemitteilungen/DE/2026/Pressemitteilung-vom-20-05-2026.html

Federal Government of Germany
Strategy on China of the Government of the Federal Republic of Germany, 2023.
www.auswaertiges-amt.de/blob/2608580/810fdade376b1467f20bdb697b2acd58/china-strategie-data.pdf

Federal Office for Economic Affairs and Export Control
Export control and academia; guidance on research cooperation, visiting researchers, publications and Dual-Use control duties.
www.bafa.de/DE/Aussenwirtschaft/Ausfuhrkontrolle/Academia/academia.html

Federal Office for Economic Affairs and Export Control
Export Control and Academia. Guidance for Science and Research, 2025. www.bafa.de/SharedDocs/Downloads/DE/Aussenwirtschaft/afk_aca_handreichung_wissenschaft.pdf

European Union
Council Recommendation of 23 May 2024 on enhancing research security.
eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=OJ:C_202403510

SIPRI / EU Non-Proliferation and Disarmament Consortium
The EU Research Security Initiative: Implications for the Application of Export Controls in Academia, 7 March 2025.
sipri.org/sites/default/files/2025-03/eunpdc_94.pdf

DAAD
Academic cooperation with China: a realistic approach, 2024.
static.daad.de/media/daad_de/pdfs_nicht_barrierefrei/der-daad/240307_daad_perspektive_china_en.pdf

German Rectors’ Conference
Guiding questions on university cooperation with the People’s Republic of China, 9 September 2020.
www.hrk.de/resolutions-publications/resolutions/beschluss/detail/guiding-questions-on-university-cooperation-with-the-peoples-republic-of-china/

Joint Committee of DFG and Leopoldina
Publications on security-relevant research and Dual Use.
sicherheitsrelevante-forschung.org/thema/publikationen

U.S.–China Economic and Security Review Commission
China’s Cyber Capabilities; information on cyber operations, MSS, civil-military fusion and use of civilian IT resources, 2022.
www.uscc.gov/sites/default/files/2022-11/Chapter_3_Section_2--Chinas_Cyber_Capabilities.pdf

Center for Security and Emerging Technology, Georgetown University
China’s PLA Challenges and Competitions; information on PLA partner networks across academia, military and industry, 28 April 2026.
cset.georgetown.edu/article/chinas-pla-challenges-and-competitions

Air University / China Aerospace Studies Institute
China’s Military-Civil Fusion Strategy; analysis of links between civilian and military innovation structures, 2020.
www.airuniversity.af.edu/Portals/10/CASI/documents/Research/Other-Topics/2020-06-15%20CASI_China_Military_Civil_Fusion_Strategy.pdf

Reuters
Chinese researchers develop AI model for military use on back of Meta’s Llama; example of military use of openly available AI models by Chinese research actors, 1 November 2024.
www.reuters.com/technology/artificial-intelligence/chinese-researchers-develop-ai-model-military-use-back-metas-llama-2024-11-01/