AI Safety Hong Kong (AISHK)

1. Description
   AI Safety Hong Kong (AISHK) is Hong Kong's first (and so far, only) dedicated AI Safety fieldbuilding organisation.

   Although there are growing concerns in Hong Kong around navigating increasingly advanced AI systems, most of the government's efforts seem dedicated towards nurturing the capabilities race. Awareness about frontier AI risks seems quite neglected, and the AI safety ecosystem has very little presence in Hong Kong, despite its position as the financial heart of the Chinese frontier AI industry.

   AISHK brings AI safety into public awareness through the following:-

• Convening: we host public talks, reading groups, and socials that bring together a wide range of people (e.g.,students, researchers, technologists, lawyers, policymakers, and finance professionals, etc).

• Training: we design, organise and facilitate workshops, courses and other training and educational programmes that upskill and inform a wide range of HK stakeholders on issues of AI safety & governance.

• Research: we collaborate with local researchers, produce Hong Kong-relevant analyses and contribute to public policy discussions.

  We began in early 2026, and have so far been operating with zero funding, with the founders investing personal time, money and resources into scaling the organization.

2. What we have built:-

a) Webinars
We run a monthly webinar series. So far, we have hosted Tzu Kit Chan of Atlas Computing, Anson Ho of Epoch AI, and Prof. Sijuade Animashaun of City University of Hong Kong. Our upcoming webinars for the rest of the year will include speakers affiliated with various other well-known AI Safety orgs, like Anthropic, Future of Life Institute, the Oxford Martin AI Governance Initiative, HKU, and more.

A running list and description of webinars can be found on our Luma calendar here: https://luma.com/aishkspeakers?period=past

b) Reading groups
We run monthly in-person reading groups and socials. These are consistently oversubscribed (events routinely receive over than 60 registrations for roughly 25–30 places). These events draw a broad mix of attendees (including students, researchers, startup founders, engineers, lawyers, civil servants, finance professionals, etc). The vast majority of our first-time attendees start out unaware about AI safety as a dedicated field, and often express interest in getting more involved once they attend their first reading group.

A running list and description of our reading groups can be found on our Luma socials calendar: https://luma.com/calendar/manage/cal-5DkGqDj8Pl7e6yC?period=past

c) Training programmes
We have been appointed as logistics partner for ML4Good and TARA, two established AI safety training programmes running their first Hong Kong cohorts in 2026. We are providing local operational support and outreach for the programmes. Over time, we to introduce HK to more high-quality AI safety training, and we would like to eventually launch our own similar programmes.

d) University talent pipelines
We are seeding student-run AI Safety groups, beginning at the Chinese University of Hong Kong (CUHK). We have also attracted interest from other elite HK universities. We intend to nurture these groups into a local talent pipeline for an AI safety ecosystem in Hong Kong.

e) Research and policy
We are building a research and policy programme to bring neglected frontier AI risks into Hong Kong’s public policy agenda. E.g., we are currently drafting a white paper responding to the AI-relevant parts of the Hong Kong Government’s Five Year Plan consultations.

f) Community growth
The community growth we have spearheaded in Hong Kong is our biggest success story. We are steadily building a core group of supporters across influential Hong Kong institutions and industries. Many have reached out to ask how they can contribute to AI safety. We have referred several community members to AI safety education and upskilling opportunities (e.g., certification courses and fellowships), while others have introduced us to influential stakeholders.

Our social media channels are growing organically through event attendance, without paid outreach or marketing. For example, our Substack has grown to 300+ subscribers and has a 55% open rate, purely on the basis of our reading groups and webinars.

3. Who's Involved:
   View our team page here: https://www.aisafetyhk.org/who-we-are
4. Concrete output:
   Funding will support our operations and growth over the next year, allowing us to build on the strong early momentum we have achieved through self-funded efforts.