China's AI Security Market: A Sovereign Stack, Party Standards, and the Same Threats as Everywhere Else

Analysts at Raft have published a detailed breakdown of China's AI security market on Habr. The industry looks closed from the outside, but between 2024 and 2025 it grew to a level of maturity comparable to American and Israeli cybersecurity players.

Why do we hear so little about China's AI security?

Roughly half of Chinese AI security vendors do not even have an English-language website, which is standard for a market that serves only domestic customers. This opacity creates a false impression that China is lagging: in reality, a full-fledged AI protection ecosystem has formed there, with products covering all the key areas.

According to Raft, the market spans three categories of solutions:

• Guardrails: policy and filtering systems on the input and output of a language model
• Monitoring platforms: continuous tracking of the behavior of AI applications in production
• Agent protection: solutions for systems that do more than work with text: they call tools, access corporate APIs, and read files

"From the outside, China's AI security market is not transparent: half of the products do not even have an English landing page.

But on the inside, a full-fledged industry has grown over the past couple of years, one that is close in maturity to American and Israeli security players," the authors of the Raft study note.

Same threats, different context

The threat model in Chinese AI security is virtually identical to the global one. The Raft study identifies four main attack vectors against AI systems:

• Prompt injections: input manipulation that pushes a model to act outside permitted boundaries
• Data and secret leaks: unintentional disclosure of confidential company information through model responses
• Misuse of tools: an agent calls an API or performs actions not envisioned by the business logic
• Poisoned agent skills: compromise of plugins or components of an agent system through the supply chain

The fundamental difference is context. Chinese AI security products are built on top of a sovereign technology stack: domestic hardware, operating systems, and cloud infrastructure with no dependence on American components. Integration and certification of security solutions take place under conditions that Western vendors cannot access.

Why does the Party need its own interpretation of safe AI?

In China, the understanding of "safe AI" is defined not only technically: directives from the CCP set mandatory requirements that go beyond standard cybersecurity. These include content filtering on politically sensitive topics, alignment with official narratives, and ideological compliance.

According to Raft, this regulatory pressure has become one of the main drivers of market growth: companies are required to meet state standards, and vendors offer ready-made tools to do so. This largely explains why China's AI security market matured noticeably faster than in countries without strict state regulation of AI.

"The task of protecting AI in China is exactly the same as everywhere else.

The difference lies in the context: a sovereign hardware and OS stack, a Party-defined understanding of safe AI, and an entire ecosystem of products we have not even heard of," Raft states.

What this means

China's AI security market is not a lagging counterpart of the Western one but a parallel, mature industry with its own products, standards, and regulatory logic. As agentic AI systems spread through the enterprise sector worldwide, protecting them becomes equally urgent in every jurisdiction. For security teams outside China, the Chinese experience, especially in agent protection and the sovereign AI stack, may offer non-trivial insights.