KU Leuven | Internship at the AI Safety Lab

Fellows join a multidisciplinary cohort of five and work on a project of their own, embedded in the Lab’s day-to-day research life.

The Lab’s work is anchored in philosophy but explicitly multidisciplinary: we welcome applicants from law, political science, economics, psychology, computer science and adjacent fields.

Research programmes

Every application must indicate one of the following.

• Theories of Victory and Viable Endgames. Positive futures: what a good outcome from advanced AI concretely looks like, and what would have to be true for us to reach it.
• Democratic Control over AI Companies. AI-enabled power concentration in frontier AI companies, and the institutional, legal and political mechanisms through which democratic societies could exercise meaningful control over them.
• Other. Strong proposals outside these two programmes will be considered, provided the contribution to AI safety is clear.

What you get

• Three months in residence at the Lab’s offices in Leuven, with a desk and access to KU Leuven’s libraries and facilities.
• A monthly allowance of €2,000, covering accommodation and other living costs.
• Formal status as a visiting scholar at KU Leuven — no employment contract; fellows remain responsible for their own tax and social-security situation.
• Supervision and feedback from the Lab’s core team, and the opportunity to present work in progress at internal seminars.
• The opportunity to present at the Lab’s annual conference and summer school.

What we ask

• Presence. Fellows are expected at the Leuven office regularly during the fellowship. We do not prescribe a number of days: the programme is built around a shared office, where a question can be put to a colleague on the spot rather than saved for a scheduled call. It suits researchers who want that, and remote participation is not possible.
• Output. A completed paper by the end of the fellowship, deposited as a preprint on arXiv, PhilPapers or an equivalent repository, and submitted to a target venue identified in the research plan.
• Dissemination. Fellows present their work at the Lab and actively disseminate it afterwards.
• Contribution to the Lab’s activities. Fellows assist with the annual conference and/or the summer school — both of which also give them a platform for their own work.
• Co-authorship. While not a requirement, we encourage co-authored papers.

Eligibility

PhD candidates, postdoctoral researchers, faculty, and researchers based outside academia. Applicants from outside academia should make clear that the stay serves a research purpose. We particularly encourage applications from women, minorities and underrepresented groups.

Submission

• Curriculum vitae. Max. 2 pages, including a list of publications.
• Cover letter. Max. 1 page: motivation, fit with the Lab, and what you hope to gain from the stay.
• Research plan. Max. 1,500 words, covering the research question and why it matters for AI safety; the methodology; a theory of change (what you expect the work to affect, and through what route); the journals, conferences and workshops you intend to submit to; why the project is feasible within three months; and the research programme you are applying for.

How we choose

We assess applications on:

• Quality and originality of the proposed research.
• Fit with one of the research programmes.
• The plausibility of the theory of change — academic, policy or field-building impact.
• Feasibility within three months.
• What the applicant would contribute to the Lab’s research community.

Key dates

• 21 September 2026

  Call published

• 31 October 2026

  Application deadline

• November 2026

  Interviews (shortlist)

• End of November 2026

  Successful applicants contacted

• March – May 2027

  First cohort in residence

Contact

Send your application to existentialrisk@kuleuven.be, with “Fellowship application — [programme]” in the subject line. Questions about the programme can be sent to the same address.

Source and application: https://www.aisafetylab.eu/open-calls/