On 8 July 2026 the Ministry of Human Resources and Social Security, together with the National Development and Reform Commission, the Ministry of Industry and Information Technology and the National Data Administration, issued the Opinions on Accelerating the Development of “AI + Human Resources and Social Security” Applications. The document was published on the National Data Administration’s website, dated 22 June.
This is a concrete placement of the “AI+” action in the livelihood field. The human resources and social security field spans employment, social insurance, labour relations and talent services. With large data volumes and clear, rigid demand scenarios, it is a natural testbed for industry large models and agents.
Building high-quality industry datasets and AI corpora
The document explicitly requires building high-quality datasets and AI corpora for the human resources and social security sector: provincial human resources and social security departments, together with provincial industry and information technology authorities and data management authorities, are to build regional text, image and other data resources. This aligns with the national direction on building high-quality industry datasets.
The meaning for AI companies is that entering this scenario is not only about model capability but about the ability to organise industry data. Teams that can obtain, clean and compliantly use industry data will have the edge in tendering.
What the scenario map tells you about entry points
The document attaches a full map of AI application scenarios in the human resources and social security field. For companies offering solutions, this list essentially defines the boundaries of what can be entered — employment services, social insurance administration, labour-relations governance, and talent evaluation.
Three compliance boundaries to clarify first
Human resources and social security data contains large amounts of personal information, some of it sensitive personal information and public credit information. To participate in these scenarios, companies must first clarify at least three boundaries.
First, the legal basis for processing personal information. Information in employment and social insurance scenarios is mostly necessary for performing statutory duties or providing public services; as a technical service provider, a company usually acts as an entrusted processor and must define the purpose, scope and security measures with the entrusting party.
Second, cross-border data transfer. If the service involves cross-border elements, it must be checked against the scope of security assessment, standard contracts or certification. The CAC’s question-and-answer on cross-border data security policies published in July 2026 further explains notification-and-consent duties; see our Multi-Jurisdiction Compliance service page.
Third, algorithm and model-side duties. Providing public-facing AI services in the human resources and social security field may still trigger algorithm filing or generative AI service filing. See our Algorithm Filing service page.
Three points for AI companies
First, decide whether you are a “technology provider” or a “service operator” — the two differ greatly in filing and data-processing duties. Second, make the entrusted-processing agreement solid, especially clauses on purpose limitation, prohibition of secondary use and security-incident notification. Third, watch the pace of local dataset building; teams that participate early in standards and specification work are more likely to obtain scenarios.
For support on entrusted-processing agreements, cross-border arrangements and filing duties, see our Data Compliance and Commercial Contracts services.
This article is based on the Opinions on Accelerating the Development of “AI + Human Resources and Social Security” Applications issued on 22 June 2026 by the Ministry of Human Resources and Social Security, the National Development and Reform Commission, the Ministry of Industry and Information Technology and the National Data Administration. It is for reference only and does not constitute legal advice.