From intelligence to agency: AI’s new frontier in international security

In person (New York) and online
31 October 2025
RSVP
From intelligence to agency: AI’s new frontier in international security

UNIDIR and the Friedrich-Ebert-Stiftung (FES) are pleased to invite you to a side event to the First Committee of the UN General Assembly titled “From intelligence to agency: AI’s new frontier in international security”.

Agentic AI refers to artificial intelligence systems designed not merely to process or analyze information, but to pursue objectives and take action autonomously in order to achieve those objectives. These systems operate with a degree of independence, adaptability and initiative that distinguishes them from more conventional forms of AI. The term “agentic” highlights their function as agents: entities that exist within an environment, assess it, make decisions based on internal goals, and execute actions without requiring direct human instruction at every step. For policymakers, the crucial point is not the underlying technical architecture, but the fact that these systems can act independently and interact with each other.

The emergence of agentic AI represents a fundamental evolution from systems that provide intelligence to those that exercise agency, transforming AI from a tool that supports human decision-making to one that can act independently in pursuit of objectives. This shift carries both promise and peril. On the positive side, agentic AI could enhance the speed, precision, and resilience of certain operations – outperforming humans in managing complex, high-tempo environments where cognitive overload is likely. Systems operating at “machine speed” can make and execute coordinated decisions faster than human operators can observe and respond, potentially providing decisive advantages in contested environments. On the negative side, it exacerbates risks to human oversight, accountability, and compliance with international law, particularly when actions escalate or when attribution is opaque.

To further investigate the issue, UNIDIR’s Security and Technology Programme and the FES New York office are organizing this side event to further analyze the potential impact of agentic AI on international peace and security, with a focus on enhancing the understanding of this technology and on exploring potential governance pathways.

When and where

Friday, 31 October 2025, 13:15 – 14:30 EST, Conference Room 12, UN Headquarters in New York, and online.

A light lunch will be provided to the participants attending in person. If attending online, please consult this website to find your local time.

A complete list of speakers will be communicated in due course.

Participants

We welcome all First Committee delegates, as well as members of the multi-stakeholder community including industry, civil society and intergovernmental organizations (ECOSOC status). Kindly note that UNIDIR is not able to grant access to the UN Headquarters for those who are not already in possession of the UN accreditation.

RSVP

Registration is mandatory for in-person, including those who already have the UN grounds pass, and virtual participation. To attend please register by Thursday, 30 October 2025.

Further information

Please contact sectec-unidir@un.org if you have any questions.

In partnership with