SIPRI Report: Responsible Procurement of Military Artificial Intelligence

Representative Image

Coming in the wake of the AI Impacct Summit in New Delhi from 16 to 20 February this new SIPRI report "Responsible Procurement of Military Artificial Intelligence" [Report] by Netta Goussac and Dr Vincent Boulanin examines the intersection of military procurement and responsible military artificial intelligence (AI).

As per the Executive Summary of the Report it investigates the reasons for an how states are adapting procurement processes for military AI adoption.

Significance of the legal obligations and high-level policy commitments related to responsible military AI are also highlighted

This comes even as the geopolitical environment is characterized by strategic competition and lessons drawn from contemporary armed conflicts, many militaries are under pressure to accelerate procurement, deployment and scaling of AI capabilities as per the Report.

In this environment States are challenging defence procurement processes to facilitate rapid acquisition of military AI by adapting procurement pathways, including by (a) deepening collaboration with suppliers to better match capability needs with products; (b) adopting more iterative procurement processes; and (c) trying different methods to achieve assurance about lawfulness, safety and reliability.

Determining how states integrate responsible military AI commitments into procurement processes remains challenging. Legal and policy frameworks require procurement authorities to justify the need for military AI, independently verify supplier claims, and clarify responsibility in decision-making. Adapting procurement allows states to apply these obligations, with industry collaboration helping assess capability needs, iterative processes clarifying accountability, and assurance practices improving testing and evaluation standards.

The Report has Three Recommendations as follows: -

1. Adapt procurement processes to give effect to high-level obligations and commitments to responsible development and use of military AI. The first step is to ensure that policies, procedures and practices relating to military procurement explicitly refer to any national laws, policies and commitments related to responsible military AI as per the Report.

Implementing responsible military AI requires addressing challenges like technical expertise to evaluate necessity and supplier claims, opacity of AI systems, commercial concerns, and the ongoing evolution of AI technology.

2. States should develop and publish documents articulating clear expectations for suppliers of military AI capabilities.

Military AI suppliers need clear guidance from clients to streamline procurement and meet expectations. Publicly outlining technical standards—covering equitability, bias mitigation, traceability, reliability, security, accountability, and governability—helps translate principles into practical requirements like error rates and documentation. Clear requirements allow suppliers to build compliant systems from the start and make procurement more efficient, aligning national processes with policy and legal commitments for responsible military AI.

3. States should address the responsible procurement of military AI in international policy discussions. Thus states adapting procurement processes  should be a part of the component of international military AI governance discussions to strengthen both national implementation and broader international frameworks.

What facets of the above are tracked in the AI Impact Summit in New Delhi now remains to be seen.

LOGO AI IMPACT SUMMIT