The labor scenario in China appears to be undergoing a profound change, at least in large cities. Under pressure from a record 12.7 million graduates expected by 2026 and an economy seeking new avenues of growth, the Chinese government has launched an unprecedented offensive to promote “One Person Enterprises” (OPC – One-Person Company).
The goal is clear: transform unemployed young people into one-man armies, equipped with “AI agents” capable of managing everything from marketing to customer support.
However, the enthusiasm – revealed by data indicating the existence of more than 500 thousand companies of this type already registered across the country, with the Haidian district, in Beijing, alone adding more than 2,500 new registrations in recent months – was abruptly tempered on Tuesday, March 17, by a warning from the Ministry of State Security about the cybersecurity risks of these tools.
O AI Plus National Plan
The strategy, integrated into the AI Plus National Planaims to mitigate the youth unemployment crisis through radical automation. In tech districts like Haidian in Beijing and Longgang in Shenzhen, the business model no longer involves hiring teams but setting up autonomous agents.
The system at the center of this revolution is OpenClaw (popularly nicknamed xiaolongxia or “lobster” after its name), an open source framework that allows a single founder to operate complex processes on an ongoing basis.
Regional incentives are aggressive. In Longgang district, in Shenzhen, the local government has provided investment funds of up to 10 million yuan (around 1.3 million euros) for OPC projects using AI that demonstrate viability. In Wuxi and Hefei, support includes free housing and direct subsidies for “computing power,” allowing entrepreneurs to access powerful servers to run their agents with no upfront costs.
The “Lobster” that scares the State
OpenClaw’s meteoric success brought with it scrutiny from authorities. The Ministry of State Security (MSS) has published the Safety Manual for Lobster Farming, a metaphorical title that uses the system’s popular nickname to present a technical document that exposes critical vulnerabilities in the architecture of these AI agents.
According to the MSS, OpenClaw’s autonomy – which requires read and write permissions on central systems to be able to perform tasks independently – represents an “open window” for external actors. The government warns of the risk of remote code execution and the possibility of these agents, if misconfigured or infiltrated, exporting sensitive data to servers outside Chinese jurisdiction. The warning extends to the banking sector and state agencies, where the use of unauthorized AI agents has been banned with immediate effect.
An impossible balance?
The paradox defines the current moment in technology in China: on the one hand, the existential need for innovation and job creation, but, on the other, the obsession with data security and centralized control.
Analysts indicate that the OPC model is Beijing’s response to the slowdown in the traditional technology sector. By encouraging individuals to become autonomous producers, the State reduces the burden on large companies and decentralizes the economy. Except that the MSS intervention suggests that the seemingly full freedom of OPCs will quickly be surrounded by regulatory walls.
Leave a Reply