AI Safety Asia (AISA)

What it does

Non-profit working to harmonise international governance of safe and secure AI 'the Asia way'. Delivers AI governance training to government officials and civil society — including director-level officials in Indonesia and parliamentarians in Malaysia — conducts policy research (including with Brookings on Southeast Asian AI safety governance), and runs courses such as a 10-week cooperative AI course with the Cooperative AI Foundation. Listed in UNESCO's Global AI Ethics Observatory CSO directory and the WEF AI Global Alliance.

What they ask for — in their own terms

Asian institutions — especially Southeast Asian and middle-power states — should have real leverage over how frontier AI is built and governed, before the terms are locked in by a handful of dominant actors.

What they do

Three programmes: Curate (helps governments map AI dependencies and connect to frontier-risk scientists), Inform (translates technical foresight into safeguards for negotiations), Catalyse (maintains channels between senior officials across bridge states). Listed by UNESCO as a civil society organisation and by the World Economic Forum.

Honest assessment

Small and growing; the kind of organisation that reads its own email. Strongest value if your concern needs a hearing in an Asian jurisdiction where Western AI safety organisations have no standing. UNVERIFIED: an independent aisafety.asia homepage was not confirmed — the UNESCO profile is the verified reference.

What a newcomer can do

Subscribe to the AI governance briefing newsletter; if you are a policymaker, researcher or organisation working on AI governance in Asia, email contact@aisafety.asia.

Worth knowing

Aimed at officials and researchers, not the general public — the get-involved page explicitly invites policymakers, researchers and organisations. If you are none of those, the newsletter is realistically your only route. No verified donation route.

Concerns it covers

Governance capacity and regional representation — the concern that AI safety norms are set in the West and imposed on Asia without local capability to evaluate them. Covers safety and security broadly rather than a specific technical threat.

Government channel

Real and direct in-region — trains sitting officials and parliamentarians in several Asian states, which is unusually concrete access.

Tips

no

Where they work

Asia, with emphasis on Southeast Asia and middle powers.

Size

Small; headcount not disclosed. Non-profit.

Spending

Not disclosed; no donation page found.