Head of Frontier AI Risk

Location: Remote, global, preference for Japan, Republic of Korea, Singapore or Indonesia

Engagement: Full-time

Reports to: CEO

Team: Senior individual contributor, working across AISA's research, training and convening workstreams

Why this role, now

AI capability is moving faster than the institutions meant to govern it, and the risks are no longer hypothetical: concentration of control over frontier systems and the access they grant, and the loss-of-control failures now showing up in deployed agentic systems, are live governance problems rather than forecasts.

Asian countries have an important role to play in the international governance of AI that addresses these risks — and almost nobody is working on how. Middle powers in the region hold real leverage over the frontier AI supply chain (chip fabrication, advanced memory, data centres, subsea cables) and over some of the world's largest deployment markets, but each government currently negotiates alone, with the same handful of external suppliers, on those suppliers' information. Whether Japan, South Korea, Singapore, Indonesia and others end up shaping the emerging international regime — or importing someone else's rules along with the technology — is still open.

We are hiring someone to help us figure out how they should.

About AI Safety Asia

AISA builds the governance capacity that lets Asia's middle powers shape how frontier AI is developed, deployed and governed. We run three connected activities: research, capacity building and convening. Research sets the agenda and supplies the scenarios; capacity building equips the director-level officials who draft the rules; convening puts senior decision-makers inside the problem and turns the outcome into standing mechanisms.

Since March 2025 we have delivered 21 programmes, including senior officials trained across national ministries and scenario exercises for senior officials and ambassadors in Geneva and at the Biological Weapons Convention, with partners including Brookings, Chatham House and the UK's FCDO.

Our work so far has been weighted towards domestic capacity — helping individual governments build the expertise to write and enforce their first binding AI rules. We are now expanding deliberately into regional and multilateral coordination: middle-power coalitions, Track 1.5 diplomacy, and the international processes where the rules for frontier AI will actually be set. This is where we think the marginal value is highest, and it has been largely neglected by the AI safety community to date. This role is central to that shift.

You would be joining a genuinely unusual team. Our Chief Research Officer is an AI Expert to the UN Secretary-General's High-Level Advisory Body on AI and former Head of Data and AI Ecosystems at the World Economic Forum, where he helped establish the AI Governance Alliance. Our Chief Strategy Officer set up and led the Asia-Pacific policy planning unit at Global Affairs Canada. Our Senior Director of International AI Diplomacy spent two decades as a US Foreign Service Officer, most recently as Cyber, Digital & Technology Officer in Paris and previously as Regional Cyber Policy Coordinator for East Asia-Pacific. Our COO was Global Head of Strategy at TikTok. Our Chief of Staff, Global Affairs is a fellow at the Stanford Law AI Initiative. Our Director of Biosecurity was a BWC biosecurity fellow and contributed to UK Government DNA synthesis screening guidance. Our Senior Director of Multilateral and Head of China spent nearly a decade at UNDP as Head of Digital Policy and Global Partnerships and is now an ERA AI Governance fellow working on Chinese AI policy.

Our advisors include Prof. Robert Trager (Oxford Martin AI Governance Initiative; GovAI), Prof. Aaron Maniam (former Deputy Secretary, Singapore Ministry of Communications and Information; OECD Expert Group on AI Futures) and Yi Zeng (founding director, Beijing Institute of AI Safety and Governance.

AISA is part of UNESCO's Global Civil Society Organizations and Academic Network on AI Ethics and Policy, a member of the World Economic Forum's AI Governance Alliance, an official affiliate of IASEAI, and a partner organisation of Partnership on AI.

Purpose of the role

To be AISA's authoritative voice on frontier AI risk, and to shape what Asian middle powers do about it — briefing senior government stakeholders across the region, informing our strategy on the role these countries should play in addressing catastrophic risks from AI, and improving the evidence base that regional and international AI policy is built on.

The role needs an unusual combination: the technical or governance fluency to read a system card, critique an evaluation report and reason clearly about threat models, plus the diplomatic judgement to brief a minister on what it means for their country's position — and a view on where the leverage actually sits between capitals.

What you will do

Senior government engagement and regional/multilateral coordination (~40%)

• Brief ministers, permanent secretaries, regulators, AI safety institute leadership and senior officials across Asia on frontier AI capabilities, risks and governance options, in person and in writing; translate technical developments into decision-relevant advice
• Develop and advance AISA's positions in regional and international processes — middle-power coalitions on AI, Track 1.5 dialogues, ASEAN, APEC, OECD, UN and the AI summit series — including where the interests of individual capitals converge and diverge and what a workable coalition would actually ask of each
• Produce briefing notes, option papers and technical explainers pitched at senior non-technical readers, and serve as a trusted, non-partisan advisor to governments that lack in-house frontier AI expertise
• Provide the technical substance for AISA's convenings, tabletop exercises and capacity building programmes, including scenario realism and post-exercise analysis

Doing the research (~35%)

• Lead AISA's own research on frontier AI risk and its governance: dangerous-capability evaluations and their limits, safety cases and assurance, agentic systems and autonomy, misuse pathways including cyber and biosecurity, loss-of-control risk, power concentration and lock-in, compute and supply-chain governance, incident reporting and post-deployment monitoring, and open-weight release questions
• Work out what different Asian countries can actually do about these risks, individually and together — grounding the work in regulatory capacity, adoption patterns, deployment contexts and strategic interests genuinely present in the region, and building the case for what a government needs to test, verify, refuse or coordinate on for itself
• Publish across registers (peer-reviewed, preprint, policy briefs and reports) and represent this work publicly at technical and policy venues, in the press where appropriate, and in multilateral processes

Setting direction — for AISA and for the wider field (~25%)

• Inform AISA's strategy on what role Asian middle powers should play in addressing catastrophic risks from AI: which countries, which risks, which venues, and in what sequence
• Set the technical direction of AISA's wider research portfolio and act as technical reviewer and quality bar for AISA outputs, including work from colleagues, fellows and commissioned researchers
• Advise AISA leadership on where the field is heading and what it means for our strategy, and help shape the external research agenda so that the questions Asian governments need answered actually get worked on

Who we are looking for

This is a rare combination of expertise. We would rather hear from you than not — please apply if the core of this describes you, even if you don't tick every box.

Essential

• Deep expertise in frontier AI — via either of two routes. We are open on which:

• Real technical depth in frontier AI — you understand how modern frontier systems are trained, evaluated and deployed, and can assess capability claims on the evidence; typically from research or engineering work on frontier or near-frontier systems, evaluations or safety work at a lab, institute or serious academic group; or
• Deep understanding of frontier AI governance, particularly with respect to catastrophic risk — you have worked substantively on the governance of frontier systems and can hold your own on the technical substance, for example from a policy research institute, an AI safety institute or an international process. An international relations or policy background with genuine depth on AI global catastrophic risk is as valuable to us as frontier engineering experience.

• Fluency across the threat models — you can reason precisely about loss of control, concentration of power, misuse and gradual disempowerment, distinguish them where they call for different policy responses, and say honestly which you think are most decision-relevant and why
• A view on Asia's role, not just Asia experience — significant professional experience across Asian jurisdictions, a working understanding of how AI policy is being made in the region, and a considered theory of change for what different Asian countries can contribute to addressing catastrophic AI risk. We will want to hear your actual take on questions like what role Japan or South Korea could play in nascent middle-power coalitions on AI, and where each country's leverage and priorities lie.
• Substantive policy experience — you have worked with or inside government, a regulator, an international organisation or a policy institute, and understand how policy gets made
• Command of a senior room — the presence and judgement to brief a minister and be challenged, including the confidence to say what is not known
• Excellent written English, specifically for senior officials: short, precise, and free of jargon
• Independence and self-direction, comfortable setting your own agenda in a small, distributed, fast-moving organisation
• Non-partisan integrity — commitment to honest technical advice in the public interest, including when inconvenient for AISA, industry, or the prevailing narrative

Desirable

• Experience of multilateral or minilateral processes (UN, OECD, GPAI, G20, APEC, ASEAN or the AI summit series), or of coalition-building between governments
• Published research on AI risk, evaluations, safety or governance
• Experience at an organisation working on frontier AI risk and its governance — for example an AI safety institute, a national AI regulator, a research institute of the Forethought / IAPS / GovAI type, or a frontier lab's safety, policy or evaluations function
• Working proficiency in an Asian language in addition to English
• Experience commissioning or supervising research, or building a research programme
• Track record of translating technical work for non-technical audiences

Terms and practicalities

• Structure: Full-time.
• Location: Fully remote and open globally, provided you can maintain substantial working-hours overlap with Asian time zones — preference for Japan, Republic of Korea, Singapore or Indonesia (but not essential)
• Travel: Regular regional and international travel, roughly one trip per month on average, concentrated around convenings and briefings.
• Compensation: Competitive, benchmarked to experience and location. Reasonable pre-approved expenses reimbursed.
• Term: Permanent, subject to an initial review period.

Integrity, independence and conflicts of interest

AISA's credibility depends on governments trusting that our advice is not shaped by commercial interest. Candidates will be asked to declare current and recent advisory roles, board positions, equity holdings and other engagements, including any with AI developers, so potential conflicts can be identified and managed. Our team is a diverse mix of diplomacy, private tech, NGO and labs and we see disclosed experience and interests as an asset.

How to apply

Send the below to contact@aisafety.asia with subject "Head of Frontier AI Risk — [Your Name]"

• CV
• Short cover letter — max two pages, covering why this role, what role you think Asian middle powers should play in the international governance of frontier AI risk (and where the current Asia AI-risk policy conversation is getting something wrong), and your first-six-months priorities
• Writing sample; ideally something written for a decision-maker

Process:

1. Initial conversation with AISA leadership (~45 min)
2. Technical and threat-model discussion (~45 min)
3. Scenario-based briefing exercise (~45 min)
4. References

We may also include a paid work test, this will be confirmed as part of the process.

Applications reviewed on a rolling basis, with a deadline of 27 September; informal, confidential enquiries welcome before applying.

AI Safety Asia is committed to building a team that reflects the diversity of the region we serve. We welcome applications from candidates of all backgrounds.