China's AI security industry has caught up with the US and Israel, Habr reports

China's AI Security market is opaque to outside observers — half of the products don't even have an English landing page — but over the past few years the country has built an industry that has come close in maturity to American and Israeli information security players. This is reported by RAFT in its company blog on Habr.

What problems Chinese AI Security solves

The task of protecting AI in China is exactly the same as everywhere else in the world: you need guardrails that check a model's input and output, you need monitoring systems, and you need separate protection for AI agents, which no longer just reply with text but move through tools, files and corporate APIs. This means the growth of China's AI Security market did not come from a fundamentally different framing of the problem, but from the same demand for agent security as in the West.

• Half of the products on China's AI Security market don't even have an English landing page
• Over the past few years a full-fledged industry has grown there, close in maturity to the US and Israeli markets
• Protection is now needed not only for models, but also for agents that work with tools, files and corporate APIs

What threats China is fighting

The threat model in China will be familiar to any AI Security specialist: prompt injections, leaks of data and secrets, misuse of tools, and poisoned agent skills.

"From the outside,

China's AI Security market is not transparent: half of the products don't even have an English landing page, while inside, over the past couple of years, a full-fledged industry has grown up that is close in maturity to American and Israeli infosec players," says RAFT's blog on Habr.

How it differs from the Western market

The task is similar, but the difference between the Chinese and Western AI Security markets lies in context: a sovereign hardware and operating system stack, a party-defined understanding of safe AI, and a whole ecosystem of products that almost no one outside China has heard of.

This means that even a specialist well versed in the American and Israeli AI Security markets will, on first contact with the Chinese sector, face unfamiliar product names and solutions built for a different hardware and OS stack — even though the underlying protection tasks remain the same.

What this means

The AI security industry is developing in parallel across different technology blocs. In China it has grown to a level comparable with the US and Israel, and simply remains invisible from outside because of a language and infrastructure barrier.

Frequently asked questions

Which threats are relevant to AI agents in

China and worldwide?

The same everywhere: prompt injections, leaks of data and secrets, misuse of tools and poisoned agent skills — this is how RAFT's blog on Habr describes the threat model.

How does

China's AI Security market differ from the American and Israeli ones?

The task is the same, but the context is different: a sovereign hardware and OS stack, a party-defined understanding of safe AI, and an ecosystem of products unfamiliar outside the country.

Why is China's AI Security market invisible from outside?

Because half of the products don't even have an English landing page. RAFT notes that the market is opaque to outside observers, although inside the country the industry is already mature.