China Rushes to Bridge AI Security Gap as Anthropic and OpenAI Lead with Advanced Models
China's AI Cybersecurity Market Set to Soar, but Challenges Remain
China's artificial intelligence (AI) cybersecurity industry is expected to reach a staggering value of US$8.7 billion by 2030, according to projections from IDC. This represents a more than 37-fold increase from the estimated 1.58 billion yuan in 2025. The rapid growth is being driven by the increasing adoption of AI technologies among domestic cybersecurity vendors, who are leveraging these tools to enhance their products and services.
While global tech giants like Anthropic and OpenAI continue to push the boundaries of AI with models that offer advanced cybersecurity capabilities, China is also making significant strides in its own AI-driven cyber defense market. For instance, Anthropic’s Mythos model, launched in April, has demonstrated an ability to identify and exploit cybersecurity vulnerabilities at an unprecedented speed and efficiency. This has sparked global interest and concern about the implications for traditional cybersecurity practices.
Austin Zhao, senior research manager at IDC China, noted that while China may not yet have a direct equivalent to Mythos, the overall trend toward AI-driven cybersecurity is inevitable. “The capabilities of China’s models are also rapidly increasing,” he said. However, there remains a gap between Chinese models and their global counterparts, particularly in areas such as zero-day vulnerability discovery, where cyber defenders have little time to respond before threats are exploited.
AI Models and Their Impact on Cybersecurity
Foundational AI models, like Mythos, carry major implications for cybersecurity due to their powerful "dual use" coding capabilities. These models can both discover and exploit vulnerabilities in software systems, raising concerns about the potential displacement of traditional cybersecurity vendors by AI companies. Despite this, Chinese cybersecurity vendors have shown a strong track record of integrating AI advances into their offerings.
Zhao highlighted that during the early days of AI-powered cybersecurity, several Chinese vendors invested heavily in the technology, allowing them to adapt quickly. However, challenges remain, particularly in accessing the latest advancements from U.S. tech companies. For example, Anthropic has not allowed Chinese vendors to use Mythos to patch vulnerabilities, while OpenAI has restricted access to its GPT-5.5 model to select U.S. cybersecurity firms.
This exclusion has raised questions about whether Chinese vendors can keep pace with their U.S. counterparts. Zhao emphasized that collaboration will be key for Chinese vendors to overcome these challenges. “Cybersecurity vendors will need to proactively collaborate with AI companies and industry associations,” he said.
The Rise of AI Security Products
In addition to AI-powered cybersecurity products, which focus on discovering and patching vulnerabilities, a newer sector known as AI security products is gaining traction. These tools are designed to protect AI systems themselves, including areas such as AI model security assessments and agentic AI risk detection.
As AI systems become more autonomous, the risks associated with vulnerabilities are growing. This has led to increased regulatory attention, with experts like Barbara Li, a Shanghai-based partner at Reed Smith, noting that regulators are prioritizing concerns over agentic AI. Different regulatory bodies are drafting guidance on how companies should deploy AI agents safely.
IDC forecasts that China's AI security market will reach 34.03 billion yuan by 2030, a more than seven times increase from 4.41 billion yuan in 2023. Zhao pointed out that high market interest is focused on evaluating the security of code generated by AI agents.
Regulatory Efforts and Industry Collaboration
Given the fast-changing nature of AI and the uncertainties surrounding its future, Beijing has sought to promote dynamic governance through close collaboration between industry and authorities. This model, referred to as "cooperative governance," aims to ensure that AI development aligns with national interests and security requirements.
Recent developments highlight this approach. On Friday, three of China's leading regulatory bodies—the Cyberspace Administration of China, National Development and Reform Commission, and Ministry of Industry and Information Technology (MIIT)—jointly released the country's first major policy framework specifically covering AI agents. The framework emphasizes the importance of developing "security technologies" such as cryptographic safeguards and cyberattack detection.
A day earlier, the China Academy of Information and Communications Technology, a research body under MIIT, announced a new alliance between big tech firms, including Huawei Technologies and Tencent Holdings, and institutions such as the Chinese University of Hong Kong. The goal of this alliance is to promote open-source standards to better verify and audit the behavior of AI agents.
As China continues to invest in AI-driven cybersecurity, the path forward will require not only technological innovation but also strategic collaboration across industries and regulatory bodies. The coming years will be critical in determining whether China can bridge the gap with global leaders and establish itself as a major player in the AI cybersecurity landscape.
