State Council office sets 18 tasks for employment-friendly AI

The General Office of the State Council published an opinion on October 23, 2026 on adapting to artificial intelligence in a way that keeps employment at the center. The Chinese text, document 国办发〔2026〕29号, is dated October 16 and marked for public release. An English summary from Xinhua, posted the same day on the government’s English site, says the opinion sets out 18 tasks in six areas. The text is guidance to ministries and local governments. It does not publish a budget, a fine, or a jobs target.

The opinion says the aim is to strengthen the ways AI can add jobs, raise the quality of work, and support new businesses, and to slow the replacement of jobs when AI is applied. The English summary uses the same frame: create jobs, improve employment quality, stimulate entrepreneurship, and mitigate displacement.

What the 18 tasks cover

The Chinese text groups the tasks in six blocks.

On new work, it tells eastern regions to host core AI industries and guides central and western regions to take on upstream and downstream work such as data labeling and the manufacture of intelligent terminals. Firms it describes as intelligent-native (智能原生企业), and “little giant” specialists (专精特新“小巨人”), can be brought into an existing special loan program for stabilizing and expanding jobs when they qualify under the rules. A separate line starts an AI entrepreneurship support plan for small and medium-sized firms, tied to open-source tools and to low-cost digital products, and says small firms and sole traders should have equal lawful access to data, algorithms, and other AI inputs.

Two state funds, the national venture-capital guidance fund and the national AI industry investment fund, are to support AI firms on market terms. The opinion also calls for a high-level open-source community and for startup help around AI applications and the conversion of technical results into businesses.

On occupations, it says new job categories and standards should be developed faster, and it names three titles as examples: digital managers (数字化管理师), AI engineering technicians (人工智能工程技术人员), and AI trainers (人工智能训练师). Local governments are encouraged to add matching professional-title fields and to run skill-grade certification.

On existing industries, the opinion wants human-machine collaboration in industrial upgrading and in the digital upgrading of agriculture, and more service work built around AI. It says AI should be rolled out first in posts where labor is short or the setting is dangerous. In labor-intensive industries, it says the scope, intensity, and pace of AI use should be handled carefully so the hit to workers is as small as possible. It encourages firms to explore an “unemployment buffer”: when AI is rolled out at scale, tell affected workers ahead of time and offer help moving to other work.

A risk check sits beside that. The opinion says AI projects in market-access fields that could cause a relatively large employment shock should have an employment-risk assessment at the same time as their review, and departments should line up transition help. Projects in other fields should be assessed in light of the circumstances. The text does not define “relatively large” with a number.

Training, services, and protections

The opinion tells universities and vocational schools, including technician colleges, to adjust majors toward AI, add short “micro-major” programs, and let leading tech firms help with courses and equipment. It points local training at the existing drive titled 技能照亮前程, and says project-based digital courses should track AI demand. Graduates and other young people, migrant workers, and unemployed people are named as groups to bring into AI employment and startup training. Public platforms are to carry more free courses.

Employment services are to use AI to match people with policies they qualify for, and national job-information systems are to be upgraded. The opinion also says AI should be used more deeply in mediation and arbitration of labor disputes, and it encourages hiring tools such as recruitment models, automated interviews, and virtual-reality training.

Workers the opinion treats as displaced by AI applications are to get job help under existing rules: guidance, training, referrals, and startup support. Insured unemployed people are to receive unemployment-insurance benefits in full and on time, again under the rules.

On rights, the opinion calls for study of protections suited to human-machine work, and it tells officials to press firms to meet existing legal duties on pay, rest, and workplace safety. It says AI used in human-resources management should be regulated, and that discrimination in hiring, fake job ads, and invasions of privacy or rest time that come from misuse of algorithms should be handled under the law. That is an instruction to apply the law. The opinion does not name a company or a person as having broken it.

Flexible workers and people in new forms of work are encouraged to join the urban employee pension, and the pilot for occupational-injury cover in new forms of work is to be expanded. Pay systems, the opinion says, should fit AI-sector work, including incentives tied to knowledge, and it mentions smart agents as a form of work to develop.

The last block is monitoring. Departments are to share industry and employment data, run surveys of how far AI has spread and which posts it creates or replaces, and set up AI-and-employment research centers at universities and institutes that already exist. AI-related employment risk is to be brought inside the existing system for handling large-scale unemployment risk. Human resources and social security departments, and development and reform departments, are told to coordinate. Major matters go to the Party center and the State Council through the usual channel.

Figures in the English summary

The English release adds context that is not in the opinion’s articles. It says, citing official figures, that the core AI industry was valued at more than 1.2 trillion yuan (about 178.62 billion U.S. dollars) in 2025, that more than 6,200 AI companies were operating, and that by the end of 2025 more than 30 percent of manufacturing enterprises with annual main-business revenue of at least 20 million yuan had adopted AI. Those are the release’s figures. This desk has not re-derived them. The same release also cites hiring-platform statistics. Those platform claims are not repeated here.

Why it matters (analysis)

The useful line is what kind of document this is. It is a State Council General Office opinion, agreed by the State Council, that tells the bureaucracy to weigh jobs while AI spreads. The employment-risk assessment is attached to review of a defined class of projects. The buffer period is something firms are encouraged to explore. Neither line, in the text published on October 23, sets a threshold, a penalty, or a date by which a firm must comply. Anyone reading the 18 tasks as a spending bill, or as a count of jobs to be created, is past what the opinion says.

This report is not legal, tax, or employment advice.

This report is not legal, financial, or health advice. See the disclaimer.