China’s Cybersecurity Companies Advancing Offensive Cyber Capabilities Through Attack-Defense Labs
Private-sector attack-defense labs form a core pillar of how China builds, sustains, and operationalizes cyber capability for commercial purposes and state-linked cyber operations.
Western governments are grappling with how private-sector offensive cyber capabilities should fit into state operations. This raises a number of practical questions: If a state tasked a company with carrying out cyber operations against an adversary, who inside those organizations would actually carry out offensive work?1 How would these units be structured for government tasks? And how would offensive activity coexist with a company’s day-to-day R&D and commercial operations?
In China, these questions are far less abstract. Private companies have been core contributors to national cyber capability building for years, supported by both policy and institutional design. They develop many of the tools, techniques, and forms of expertise that underpin defensive security products and can also be leveraged for state-sponsored cyber operations. The clearest organizational expression of this approach is companies’ widespread use of attack-defense labs (攻防实验室), internal units that merge defensive research, offensive experimentation, live-fire exercises, and product development for commercial needs and, at times, state-linked activity. The Natto Team has addressed this phenomenon (here and here), along with the underlying defense-through-offense mindset.
In 2024 and 2025, the United States sanctioned the cyber-range firm Integrity Tech (北京永信至诚科技有限公司) and the cybersecurity firm Sichuan Silence (四川无声信息技术) for conducting offensive operations, with activity in both cases traced back to their attack-defense labs, where research and innovation for their commercial products also take place. Meanwhile, Integrity Tech’s lab and Qi An Xin’s (奇安信) elite vulnerability research team Pangu Labs (盘古团队) have conducted activities in support of law-enforcement operations.
The present article examines how China conceptualizes the private sector’s role in offensive and defensive capability development, how corporate attack-defense labs function within this system and are supported by policy and market forces, and how these labs assist state cyber operations.
The Chinese Understanding of Offensive and Defensive Capability and the Role of the Private Sector
The Chinese government views offensive and defensive cyber capabilities as interdependent and part of a holistic approach to national cyber strength. While China has not openly articulated an offensive cyber strategy, official doctrines repeatedly stress that both offense and defense are essential to achieving the goal of becoming a “cyber superpower” (网络强国). These documents outline the broad structure of China’s offensive cyber forces and underscore the significant role expected of the private sector.
An early articulation of China’s approach appears in the 2013 edition of the Science of Military Strategy, published by People’s Liberation Army (PLA) Academy of Military Science – the PLA’s highest-level research institute. The document argues that the ubiquity of the internet and the blurring of civilian–military networks have expanded the number of actors engaged in cyber operations. Alongside dedicated military cyber units and relevant government departments, such as the Ministry of Public Security (MPS) and the Ministry of State Security (MSS), the document explicitly includes civilian forces, including private companies, as part of the teams that conduct cyber attack and defense. This framing has pushed private firms to develop offensive and defensive tooling and to conduct cyber operations.
To improve offensive and defensive capabilities, talent is crucial. The Chinese Ministry of Education has taken the lead in defining “attack and defense capabilities” (攻防能力) within a broader concept of live-fire (实战化) capabilities.2
In 2022, China’s Ministry of Education published a “White Paper on the Live-Fire Capabilities of Cybersecurity Talents: Attack and Defense Live-Fire Capability Edition” (English translation here). It defined four types of cybersecurity talent live-fire capabilities: “attack and defense live-fire capabilities,” “vulnerability mining capabilities,” “engineering development capabilities,” and “combat effectiveness evaluation capabilities.” Regarding attack and defense live-fire capabilities specifically, the White Paper stated that they refer to:
the ability to use cybersecurity technologies and tools to carry out security monitoring and analysis,
risk assessment,
penetration test event research and judgment [i.e. evaluation],
security operations and maintenance, and
emergency response in real business environments.
Although framed as defensive, labeling these skills as “attack and defense” live-fire capabilities implies their applicability in “attack” scenarios as well. Unlike in Western discourse, which traditionally distinguished between offense and defense, China treats the two as mutual elements of an active defense approach, with an emphasis on preemptive offensive actions and integrated offense-defense. This is not just a conceptual difference; it shapes how capabilities are funded, developed, and used in practice.
Corporate Attack-Defense Labs Among China’s Technology and Cybersecurity Giants
China’s cybersecurity industry, as a key force in building its offensive cyber capability, has taken the holistic approach that integrates both attack and defense as a core principle and central elements of its product offerings. Chinese security experts often reference the phrase “以攻促防”, meaning “use offense to promote defense.” In practice, Chinese cybersecurity companies frequently operate dedicated “attack and defense labs” (攻防实验室), where offensive and defensive research and experimentation are merged.
Over the past 15 years, China’s major technology and cybersecurity firms – including Alibaba (阿里巴巴), Baidu (百度), Tencent (腾讯), Qihoo 360 (奇安信360), Knownsec (知道创宇), Qi An Xin (奇安信), DBAPP (安恒信息), and Sangfor (深信服) – have built prominent labs that combine offensive research, defensive engineering, and live-fire experimentation. For example, Sangfor’s 6+1 laboratory forms the company’s core offensive-defensive capability system, designed to ensure it can both attack and defend effectively. At its launch, Sangfor echoed President Xi Jinping’s words: “the essence of cybersecurity work is confrontation, and the essence of confrontation lies in the contest of capabilities on both the offensive and defensive ends.” Other companies adopt similar models. DBAPP’s Starfire Lab (安恒星火实验室) focuses on practical attack-defense research and innovation, including live-fire exercises, attack simulation, threat analysis, and vulnerability discovery (see below).
Some of these large firms operate extensive networks of specialized labs: Qihoo 360 runs 19 labs, Tencent operates between 7 and 10, and Qi An Xin maintains 15. These labs focus on diverse areas such as backdoor detection, threat intelligence and attribution, software and supply-chain vulnerability discovery and exploitation, industrial control system security, and vehicle security.
Boutique Security Firms and High-End Attack-Defense Labs
Alongside China’s tech and cybersecurity giants, a parallel ecosystem of specialized attack-defense firms has taken shape. These firms operate their own attack and defense labs that focus on various types of live-fire capability research, as defined previously. Many were created in the early to mid-2010s by former employees of large technology firms and later drew investment or acquisition interest from them.
Cyber Kunlun (北京赛 博昆仑科技有限公司) is one of the most prominent examples. Widely recognized alongside Pangu Labs as China’s leading white hat team, it spun out of Qihoo 360 in 2021. Its founder, Zheng Wenbin, previously created Qihoo’s elite 360 Vulcan Team and served as the company’s CTO, while its CTO, Chen Xuebin, formerly directed Qihoo’s Vulnerability Research Institute. Cyber Kunlun now operates Kunlun Lab and the Kunlun Advanced Offensive and Defensive Team, conducting zero-day research across desktop, mobile, cloud, virtualization, and IoT systems to develop security tools.
Pangu Labs, historically known for its iOS jailbreaking research,3 exemplifies the dual nature of attack-defense labs. Its leader has argued that the essence of jailbreaking is “using offense to promote defense.” In 2021, Pangu was acquired by Qi An Xin, one of China’s largest cybersecurity companies.4 Another example is Sichuan Silence, which hosts the Double Helix Attack and Defense Laboratory. The company was once part of Topsec (天融信), one of China’s earliest and most influential cybersecurity firms and the developer of China’s first indigenous firewall.
Many of these labs, including those operated by Pangu,5 Qihoo 360,6 Cyber Kunlun,7 and Sichuan Silence,8 have demonstrated their capabilities on a global stage through high-profile hacking competitions and U.S. bug-bounty programs over the past decade.9 Their strong performance has reinforced their technical credibility, driven continued investment, and increased government interest in leveraging their expertise for state cyber operations.
Growing Market Demand and Policy Support
Collectively, attack-defense labs constitute the backbone of corporate-level offensive and defensive capability building in China. Industry trends suggest the demand for attack-defense capability is expanding.
According to a December 2023 analysis by Anquan 419, a Chinese cybersecurity industry news outlet, China’s cybersecurity market is shifting from a compliance-dominated model to one increasingly focused on attack-defense. A January 2025 report published by Hufu Think Tank (虎符智库), founded by Qi An Xin, further supports this trend, noting that attack-defense live-fire capability is increasingly regarded as the primary engine of cybersecurity innovation.
Recent policy developments reinforce the trajectory. A July 2025 draft regulation titled “Capability Requirements and Evaluation Specifications for Assessment Organizations of Classified Protection of Cybersecurity (网络安全等级保护测评机构能力要求和 评估范围),” which closed for comments in September 2025, requires “assessment organizations” – that is, certified cybersecurity testing firms – to maintain a dedicated attack-defense lab (专门的攻防实验室) with at least 10 technicians. These technicians must have experience in national-level security monitoring, threat warning, drills, and emergency response, and be able to integrate offensive and defensive skills, conduct technical research, and cultivate talent. This represents an update to the 2018 requirements, which did not explicitly mandate the creation of attack-defense labs.
Chinese attack-defense labs have also been in demand internationally. Tencent’s Keen Lab – known for its automotive cybersecurity research10 – received the 2018 “BMW Group Digital and IT R&D Technology Award” for uncovering flaws in models produced by BMW, a leading German automaker. BMW Group reported that it was discussing options for “joint in-depth research and development activities” with Tencent Keen Lab, though we could not identify more recent examples of collaboration. Keen Lab also conducted an eight-month code audit with Mercedes-Benz, another major German automaker, releasing the “Mercedes-Benz Automobile Information Security Research Review Report” in May 2021, which disclosed several major vulnerabilities, including four enabling remote code execution.
Attack and Defense Labs Assist State Cyber Operations
Chinese government agencies have increasingly drawn on the growing capabilities of attack-defense labs within domestic cybersecurity firms to support state cyber operations with offensive tool development, vulnerability research and exploitation, and counter-cybercrime support.
A first set of examples illustrates how attack-defense labs contribute to operational offensive tooling. Integrity Tech, one of China’s leading cyber-range developers, was linked to the Flax Typhoon threat actor. According to a 2024 FBI affidavit, the group used an application called Sparrow (麻雀) to control a botnet capable of file transfers, remote command execution, lateral movement, and cyberattacks. The Sparrow repository, along with embedded tools, repeatedly referenced the “KRLab,” Integrity Tech’s internal attack-defense lab.11 A similar pattern emerged in November 2025, when MSS-affiliated clusters including UNC5174 were linked to the dual-use VShell remote access tool (RAT). European cybersecurity firm NVISO assessed that DBAPP Security’s Starfire Lab maintained VShell in a 2025 report. The report noted that VShell began as a defensive research project around 2021 but later evolved into a functional backdoor used by several China-nexus threat actors. How the tool moved from DBAPP’s lab into operational use remains unclear.
Another set of cases illustrates that beyond China’s mandated vulnerability disclosure system – which channels new findings to government authorities – some labs also appear to develop exploit code. In October 2024, Sophos’ “Pacific Rim” investigation observed devices running commands consistent with exploit development and concluded that they belonged to Sichuan Silence’s Double Helix Research Institute, an attack-defense lab. In the same period, a Financial Times investigation drawing on intelligence in a White House national-security memo reported that Alibaba employees “had transferred knowledge about ‘zero-day’ exploits to the PLA.” Although the report did not specify the internal Alibaba unit involved, the reference to zero-day vulnerability or exploit work could point to Alibaba’s Orion Lab (猎户座实验室), known for conducting vulnerability research in U.S. products.
Other cases show attack-defense labs contributing to law-enforcement operations to tackle cybercrime. Pangu Labs has supported investigations by exploiting vulnerabilities and using targeted phishing to infiltrate suspects’ systems. Integrity Tech’s KRLab similarly claimed to provide long-standing cybercrime intelligence and investigative support to national and provincial public security organs, helping resolve hundreds of cases.
Defense Builders or Operation Enablers?
China’s cybersecurity industry has evolved around the idea that offense and defense develop together. This mindset has shaped how companies organize, how talent is trained, and how capabilities ultimately flow into state activity. Attack-defense labs sit at the center of this system: they drive commercial innovation while generating tools, techniques, and expertise that both strengthen defensive capabilities and enable state-sponsored operations.
For China, this integrated model is now a structural feature of the cybersecurity market, reinforced by policy, investment, and the broader commercial ecosystem in which these labs operate. Natto Thoughts has conducted extensive work to understand this ecosystem and how these companies’ capabilities intersect with state interests (see here, here, and here). More research is still needed to understand how these labs function in practice and for assessing how China’s broader cyber capabilities and operations may evolve in the years ahead.
The term “offensive” in a cyber context has had a wide variety of definitions: from a narrow focus on disruptive and destructive activity — as Mei Danowski noted in the book chapter Becoming a Cyber Superpower: China Builds Offensive Capability with Military, Government, and Private Sector — to a broad range of “proactive defense” activities such as shutting down accounts or the use of tools such as penetration testing. For the purposes of this piece, “offensive cyber operations” refers to capabilities or actions involving unauthorized intrusions conducted without the consent of the user, owner, or operator of a network or system, not activities carried out within one’s own systems to neutralize a threat.
Natto Thoughts elucidated this as early as March 2024 to help explain the context of the i-SOON leaks. See i-SOON: “Significant Superpower” or Just Getting the Job Done?
Jailbreaks are software tools that exploit system vulnerabilities to remove restrictions on Apple devices, allowing users to install unauthorized apps and customize iOS beyond Apple’s limitations.
Qi An Xin supplies products and services to over 90% of China’s central government departments, state-owned enterprises, and major banks. It has consistently invested in firms specializing in attack-defense capabilities. For example, until 2018, it was the largest shareholder of Integrity Tech, a leading cyber range company operating five specialized security laboratories focused on advanced offensive and defensive techniques.
Beyond gaining global recognition since 2014 for its groundbreaking iOS jailbreaks, Pangu has delivered standout performances in exploit competitions such as PwnFest and the Tianfu Cup and has made notable contributions to Apple’s bug bounty program.
Qihoo 360’s labs have delivered strong performances at Canada’s Pwn2Own exploit contests and made significant contributions to the bug bounty programs of Apple, Microsoft, and Google (Android), particularly between 2017 and 2020.
Cyber Kunlun members excel in offensive and defensive security research and have participated in various national and enterprise-level attack-defense exercises across industries like energy, banking, and media. Their primary focuses encompass enterprise security, host security, and application security.
Sichuan Silence’s Double Helix research lab has a rich history of participating in hacking competitions, particularly capture-the-flag (CTF) competitions, in China since at least 2014. Its team “Sichuan Silence PKAV team” was one of 12 top teams that qualified to compete in the international hacking competitions 0CTF and XCTF Shanghai in 2016 and won the first place in total score for the Asian region and the world number 1 in the Web challenge.
Other prominent labs are described in detail in the report “From Vegas to Chengdu” by the Center for Security Studies at the Swiss Federal Institute of Technology Zurich (ETH Zurich).
Between 2015 and 2020, the Keen Team reported vulnerabilities in vehicles from Mercedes-Benz, Tesla, Toyota, and BMW.
An archived 2023 version of Integrity Tech’s website described the KRLab as focused on “engineering technological achievements for vulnerability discovery in big data platforms, IoT, and virtualization platforms,” and emphasized AI-driven offensive and defensive research. References to KRLab were removed from the website in 2023.
Thanks for writing this, it realy clarifies a lot. You've definitely nailed the fundamental difference in how they operte there. Do you think this blurring of lines is sustainable for the private sector in the long run?
Very well done. I didn’t realize just how integrated their private cybersecurity sector is with the state apparatus.