# AI training for HR teams in Hong Kong: what to cover and where to start > Learn what to cover in AI training for HR teams in Hong Kong, including data rules and how it differs from a certificate course with a real case study. - URL: https://hyperfocusam.com/blog/ai-training-for-hr-teams-hong-kong/ - Language: en - Author: Sam Wong (黃力桐), https://hyperfocusam.com/ - 中文版: https://hyperfocusam.com/blog/ai-training-for-hr-teams-hong-kong-tc/ (Markdown: https://hyperfocusam.com/blog/ai-training-for-hr-teams-hong-kong-tc/index.md) - Published: 2026-10-30; revised 2026-10-30 --- **Short answer:** good AI training for an HR team starts with HR's own work, not a tool tour: job descriptions, interview questions, policy drafts, onboarding packs and survey comments. It starts with data rules, because HR holds more personal data than almost any other team. And it often comes first in the company, because HR then helps everyone else adopt AI. *Sam Wong (黃力桐) is a corporate AI trainer based in Hong Kong. 11,000+ people trained, 70+ organisations, 13 countries.* ## What should AI training for an HR team cover? The tasks HR repeats every week, done with the AI tools the company has already approved. In my sessions that usually means: - **Writing:** job descriptions, job ads, offer and rejection letters, internal announcements. - **Hiring:** structured interview questions for a role, and a scoring guide the panel can share. - **Policy:** a first draft of a policy or an FAQ for staff, written from the company's own rules. - **Onboarding and training:** a first-week pack, a short quiz, a session outline. - **Listening:** grouping open comments from an engagement or exit survey into themes, after the comments have been stripped of names. Every one of those tasks ends with a person checking the output. AI drafts; HR decides. ## Why does data safety come first for HR? Because HR handles CVs, salaries, medical leave and performance notes. Before anyone opens a tool, the team needs a simple rule for what may go in and what may not. I use a red, amber and green check: green is public, amber needs care, red never goes into an AI tool. The company's own policy always comes first. I wrote the full version in [can staff put company data into AI?](https://hyperfocusam.com/blog/staff-ai-data-rules-hong-kong/) ## Why does HR often go first? In the programmes I have run, HR is often the first team to train and then the team that carries the rollout. At Garden, the Hong Kong food manufacturer, the programme started with HR and then moved to sales and marketing, manufacturing operations, finance, purchasing, R&D and senior management. At a major Hong Kong bank, the programme was run by the bank's HR learning and talent development centre and opened to all staff. When HR has used AI on its own work first, it can plan the next cohorts from experience rather than from a brochure. I made the same point to HR professionals at a [Hong Kong Productivity Council seminar](https://hyperfocusam.com/en/cases/hkpc-registered-speaker/) in May 2026: first AI literacy for all staff, then teams redesigning their own workflows, then company rules for how people and AI agents work together. ## How is in-house training different from an HR AI certificate course? Public certificate courses are useful for one HR professional who wants a broad grounding. In-house training is built for your team: | | Public HR AI course | In-house training for your HR team | |---|---|---| | Who attends | Individuals from different companies | Your HR team together | | Tools | Whatever the course uses | The tools your company has approved | | Examples | General HR cases | Fictional versions of your own tasks | | Data rules | General principles | Your own data policy, practised in the room | | After the session | You apply it alone | The team agrees on shared ways of working | ## What did this look like at Garden? [Garden's department-by-department programme](https://hyperfocusam.com/en/cases/garden-ai-pioneer-programme/) began with HR. The company's IT rules allowed only Microsoft Copilot on the web and Perplexity, and the most confidential tier of data was barred from AI tools altogether. So every exercise used fictional data, and every session opened with the data-security card. The results are on the case page: after the first cohort, the company commissioned three further rounds of training in 2026. Seven of the 19 colleagues in the first cohort volunteered to champion AI in their own departments. Across 2026 the programme ran 20 in-person sessions for 7 cohorts. ## What should HR prepare before the first session? 1. **The approved tools list,** and who has a paid licence. Versions often differ between colleagues. 2. **The data policy,** even if it is short. 3. **Three to five real tasks** the team repeats every week, so the exercises can copy their shape with fictional data. 4. **A name or two** who might become AI champions afterwards. For how I run training for teams across a company, see [corporate AI training in Hong Kong](https://hyperfocusam.com/corporate-ai-training-hong-kong/) ([中文](https://hyperfocusam.com/zh/corporate-ai-training-hong-kong/)). ## FAQ ### What should AI training for an HR team in Hong Kong cover? AI training for HR teams in Hong Kong should focus on repetitive weekly tasks, such as drafting job descriptions, interview questions, policy documents, onboarding packs, and survey themes. Training must use the company's approved AI tools and begin with data security rules, as HR handles sensitive information like CVs and salaries. Every AI-generated output must be reviewed by a person. ### Can HR staff put employee data into AI tools? HR staff may only input employee data if it aligns with the company's specific policy. The Hong Kong Personal Data (Privacy) Ordinance applies to all information uploaded or typed into an AI tool. To ensure compliance, use a traffic-light system: public information is safe, sensitive data requires caution, and confidential personal records like medical history or salaries must not be uploaded unless explicitly permitted. ### Why should HR be trained in AI before other departments? HR teams should receive AI training first because they are responsible for planning and delivering corporate training. By mastering AI tools through their own work first, HR can design effective subsequent cohorts based on practical experience. For example, at Garden, a Hong Kong food manufacturer, Sam Wong's programme began with HR before expanding to five other functions and senior management. ### How is in-house AI training different from a public HR AI certificate course? A public course trains individuals from various companies using general examples. In-house training brings a specific HR team together to use approved company tools and practice with fictional versions of their own actual tasks. This approach ensures the team applies their specific data policies and leaves with shared, consistent ways of working tailored to their organisation. ### What should an HR team prepare before AI training? Before starting, the HR team should provide a list of approved AI tools and license holders, the company's data policy, and three to five recurring weekly tasks. Identifying one or two potential AI champions is also helpful. This preparation allows the trainer to create practical exercises that mirror the shape of real work while using fictional data. --- **Planning AI training for your HR team?** [Message me on WhatsApp](https://wa.me/85264315177?text=Hi%20Sam%2C%20I%20read%20your%20article%20about%20AI%20training%20for%20HR%20teams) and tell me a little about your team. ## Sources - Garden: department-by-department AI training: [/en/cases/garden-ai-pioneer-programme/](https://hyperfocusam.com/en/cases/garden-ai-pioneer-programme/) (read 30 October 2026). - A major Hong Kong bank, online AI series: [/en/cases/hong-kong-bank-ai-training/](https://hyperfocusam.com/en/cases/hong-kong-bank-ai-training/) (read 30 October 2026). - Hong Kong Productivity Council, talks for HR professionals and SMEs: [/en/cases/hkpc-registered-speaker/](https://hyperfocusam.com/en/cases/hkpc-registered-speaker/) (read 30 October 2026).