The first multi-disciplinary Danish conference on the existential threat from artificial intelligence.
About
The Danish AI Safety Conference convenes researchers, students, policymakers, and others concerned with the catastrophic and existential risks posed by increasingly capable artificial intelligence. It is the first national conference in Denmark dedicated to this topic, building on three smaller community-scale gatherings held in recent years.
The programme spans both technical AI safety research — alignment, interpretability, evaluations — and the governance and policy questions raised by frontier AI systems. The aim is to strengthen the Danish AI safety community and to establish a serious, recurring national venue for discussion of these risks.
The conference is held on a Saturday so that researchers, students, and others with weekday commitments can all attend, in the late-August window between summer holidays and the start of the autumn semester.
Speakers
Keynote: Rasmus Herlo, Assistant Professor, University of Copenhagen.
Confirmed: Lukas Galke Poech, Associate Professor, University of Southern Denmark — Mechanistic interpretability (45 min).
Further speakers to be announced.
Programme
10:00 Doors open · coffee and registration
10:30 Opening remarks
11:00 Lukas Galke Poech (SDU) — Mechanistic interpretability
11:45 Further technical talks (to be announced)
12:30 Lunch and structured one-on-ones
14:00 Afternoon session — governance and policy
15:30 Break
16:00 Lightning talks · sign up on the day, five minutes each
17:30 Keynote — Rasmus Herlo
18:30 Group photograph and dinner
20:00 Open discussion and drinks
Indicative schedule, subject to change.
Registration
Registration is free. Capacity is limited to 60 attendees.
The first multi-disciplinary Danish conference on the existential threat from artificial intelligence.
About The Danish AI Safety Conference convenes researchers, students, policymakers, and others concerned with the catastrophic and existential risks posed by increasingly capable artificial intelligence. It is the first national conference in Denmark dedicated to this topic, building on three smaller community-scale gatherings held in recent years.
The programme spans both technical AI safety research — alignment, interpretability, evaluations — and the governance and policy questions raised by frontier AI systems. The aim is to strengthen the Danish AI safety community and to establish a serious, recurring national venue for discussion of these risks.
The conference is held on a Saturday so that researchers, students, and others with weekday commitments can all attend, in the late-August window between summer holidays and the start of the autumn semester.
Speakers Keynote: Rasmus Herlo, Assistant Professor, University of Copenhagen.
Confirmed: Lukas Galke Poech, Associate Professor, University of Southern Denmark — Mechanistic interpretability (45 min).
Further speakers to be announced.
Programme 10:00 Doors open · coffee and registration 10:30 Opening remarks 11:00 Lukas Galke Poech (SDU) — Mechanistic interpretability 11:45 Further technical talks (to be announced) 12:30 Lunch and structured one-on-ones 14:00 Afternoon session — governance and policy 15:30 Break 16:00 Lightning talks · sign up on the day, five minutes each 17:30 Keynote — Rasmus Herlo 18:30 Group photograph and dinner 20:00 Open discussion and drinks Indicative schedule, subject to change.
Registration Registration is free. Capacity is limited to 60 attendees.
Posted on: