This side event to the 81st session of the UN General Assembly First Committee will discuss key conceptual, methodological and practical considerations for developing an AI security lexicon.
The need for a lexicon for AI has been discussed on several recent occasions, including in the informal exchanges on AI in the military domain held in June 2026.
While developing a lexicon presents technical and political challenges, many States recognize its practical utility: a lexicon can disambiguate and enhance familiarity with AI security topics across policy communities, thus promoting shared understandings.
In 2026 and 2027, UNIDIR will bring together a multidisciplinary and geographically diverse group of experts to develop an AI security lexicon. It will serve as a repository of key terminology, explaining and clarifying fundamental concepts and, where needed, outlining different definitions and interpretations. The aim is to provide a neutral, technical resource that States and the broader international community can draw on in national contexts and international discussions.
This event will consider examples and lessons from other policy domains, as well as technical and terminological challenges in developing an AI security lexicon. These include curating terms and concepts, ensuring technical neutrality across scientific communities, mitigating risks of political bias or definitional capture, and ensuring long-term applicability.
Agenda
Further details will be added in due course.
When and where
Friday, 23 October 2026 | 13:15 – 14:30 EDT | In person (Conference Room 8, United Nations Headquarters, New York).
Check your local time here.
Participants
The event is open to all First Committee delegates. Please note that participants joining us in person must have a valid visitor pass or long-term badge to enter United Nations Headquarters, as UNIDIR is unable to facilitate entry arrangements.
RSVP
For in-person participation please register here.