On 4 April, Dr Yasmin Afina, Researcher in the Security and Technology Programme at UNIDIR briefed the UN Security Council on harnessing safe, inclusive and trustworthy AI for the maintenance of international peace and security. This Arria-formula meeting was organized by Greece, France, and the Republic of Korea, and co-hosted by Armenia, Italy, and the Kingdom of the Netherlands. The briefing highlighted the opportunities and risks posed by AI and the importance of responsible governance in the security domain.
In her remarks, Dr Afina emphasized that AI is capable of advancing peace and security while simultaneously introducing grave risks if not governed responsibly. “With innovation and progress proceeding apace,” she said, “the international community, States and the wider multistakeholder ecosystem are increasingly invested in ensuring the responsible development, deployment and use of this technology.”
Dr Afina welcomed the recent adoption of UN General Assembly Resolution 79/239, which she described as a “pivotal moment” in multilateral discussions on AI beyond lethal autonomous weapons systems. She also highlighted landmark efforts such as the Responsible AI in the Military Domain summits, and the Paris Declaration on Maintaining Human Control in AI-enabled Weapon Systems, endorsed by 27 States.
As a longstanding leader in this field, UNIDIR has pioneered research at the intersection of AI, disarmament and security. Dr Afina noted the Institute’s commitment to inclusive and trusted engagement, exemplified by its flagship initiative — The Roundtable on AI, Security and Ethics (RAISE), launched in partnership with Microsoft in March 2024. The RAISE initiative recently convened its inaugural Global Conference on AI, Security and Ethics, gathering participants from across all UN regional groups and the P5 countries, with nearly 500 in-person attendees and over 1,000 joining online.
Spotlighting priorities for responsible AI
Drawing on UNIDIR’s forward-looking research, Dr Afina identified six priority areas for advancing responsible AI governance:
- Capacity-building: Efforts must be context-sensitive, inclusive, and reinforce existing capabilities rather than focus solely on the new.
- Trust: Building trust among States, between users and developers, and in AI systems themselves is critical.
- Human element: Understanding human roles throughout the AI lifecycle is essential for accountability and performance.
- AI-data relationship: Greater attention must be given to data governance, including issues like bias and digital forensics.
- Whole-of-lifecycle governance: Ethical and legal considerations should be embedded “by design” from development through to decommissioning.
- Mitigating destabilization: The potential of AI as both force- and threat-multiplier must be carefully managed.
She further highlighted the transformative potential of responsibly used AI technologies in broader peace and security contexts, such as counter-terrorism, border security, and humanitarian response. However, she also acknowledged the serious risks of cyber threats, escalation, information disruption, and misuse by non-State actors.
Cooperation at all levels
Looking ahead, Dr Afina proposed a series of actionable steps for Member States and the Security Council:
- Establish an annual Security Council session dedicated to AI and its security implications.
- Provide inclusive, independent platforms for global dialogue.
- Promote regional and sub-regional cooperation through information sharing and best practices.
- Encourage the development and regular review of national AI strategies.
“UNIDIR stands ready to assist States,” she concluded, “whether through impartial research, trusted convening, or tailored capacity-building, to ensure that no one is left behind in this critical endeavor.”
This meeting represents a growing recognition within the UN system of the urgent need to address AI’s evolving role in international peace and security, reinforcing UNIDIR’s ongoing mission to inform and support multilateral action in this space.
Read the full briefing notes here.