In April 2024 the Ministry of Human Resources and Social Security and eight other departments issued the action plan for cultivating digital talent to support the digital economy (2024–2026). It covers digital skills training, digital talent evaluation and AI talent development. The document addresses talent supply, but in practice it lands as contractual arrangements between employers and employees.
Three focus areas
Digital skills training addresses supply capacity, evaluation addresses recognition standards, and AI talent development addresses shortage areas. Together they turn “digital talent” from a loose recruitment label into something that can be trained, assessed and counted.
For HR and legal teams, two consequences follow: training spend rises, and evaluation results become linked to roles and pay. Both enter employment contracts and internal policies.
Training agreements and service periods
Where an employer funds specialised training, it may agree a service period with the employee. The length should match the investment, and any liquidated damages should be limited to the training costs actually incurred. Ordinary induction cannot be repackaged as specialised training.
AI training often involves external courses, compute resources and project practice, which makes the cost boundary hard to draw. Sign a written agreement before training begins, setting out content, cost components, service period and breach liability, and keep the cost evidence.
Scope of non-compete obligations
AI talent moves frequently, and non-compete clauses are widely used. But non-compete has a statutory scope of application and cannot cover every employee; duration, compensation and geographic or business scope must all comply with the law. Applying it to ordinary engineers raises cost and may produce an unenforceable clause.
A more workable approach combines non-compete with trade-secret protection: identify the roles that genuinely touch core technology and training data, then design obligations for those roles. For everyone else, confidentiality and exit handover procedures are sufficient.
Classification risk in flexible work
Project-based hiring, outsourcing and internships are common in AI. The nature of an employment relationship is determined by actual management practice, not by the label on the contract. If a company imposes attendance, performance review and day-to-day direction on outsourced staff, a de facto employment relationship may be found. Writing management boundaries into the outsourcing contract and keeping acceptance records is the basic mitigation.
Titles and skills evaluation
Where evaluation results feed pay, promotion or public marketing, companies should verify the accreditation of the assessing body and the accuracy of certificate wording. Claims such as “national” or “officially certified” need verifiable support.
For training, non-compete and outsourcing agreements, see our commercial contracts service; for protection of core technical secrets, see IP protection. Definitions are in the glossary.
This article is based on the action plan for cultivating digital talent to support the digital economy (2024–2026), issued by the Ministry of Human Resources and Social Security and eight other departments in April 2024, and on public information. It is for reference only and does not constitute legal advice.