CyCon 2026 Series – Governing Responsible Military AI Through Defense Procurement

Editors’ note: This post introduces a series derived from panels and discussions that took place in 2026 at the 18th annual International Conference on Cyber Conflict (CyCon) event hosted by the NATO Cooperative Cyber Defence Centre of Excellence in Tallinn, Estonia. This year’s theme was “Securing Tomorrow.”

Governments around the world are developing legal and political frameworks for the use of military artificial intelligence (AI). International organizations, expert groups, and States continue to debate new norms, guiding principles, and governance mechanisms intended to ensure that AI is developed and employed in compliance with international humanitarian law (IHL). These initiatives are necessary and valuable. Their focus, however, is on future governance frameworks rather than the instruments governments already have at their disposal. While such international and national frameworks are still outstanding, States should direct their attention to the governance instruments that they already possess.

One such governance instrument is the obligation under Article 36 of Additional Protocol I to the Geneva Conventions (AP I) to review new weapons, means, and methods of warfare. Another lever, far less discussed, is defense procurement. While weapons reviews assess whether a capability may lawfully be used, procurement can influence how that capability is designed in the first place.

This article argues that governments should use defense procurement not merely as an acquisition process but as an instrument of governance. Procurement gives States the opportunity to translate legal and policy objectives into concrete technical and contractual requirements, thereby influencing the design of military AI even before deployment. Additionally, effective governance also requires complementary governance within industry. This post therefore distinguishes between external governance, exercised through regulation and defense procurement, and internal governance, which should be exercised by defense companies through integrated organizational structures.

Defense Procurement as a Governance Instrument

Defense procurement is a means of acquiring military capability. It should, however, also be understood as a powerful instrument through which governments can shape the development of military AI.

Every procurement decision requires governments to define operational use cases, capability requirements, technical specifications, quality standards, and legal requirements, which the procured solution must meet. These choices influence not only which products are ultimately acquired but also which characteristics defense companies have commercial incentives to develop. Procurement, therefore, influences technological development long before a capability reaches operational service.

This is particularly important because many of the decisions that determine whether military AI can be used consistently with IHL are taken during the system design phase rather than deployment. Software architectures, training data, assurance measures, testing methodologies, and mechanisms for human oversight are established early in development and are often difficult and costly to modify later. Governments wishing to influence these decisions should therefore do so before contracts are awarded by integrating relevant legal and ethical requirements into selection criteria within the tender process. Procurement thus enables governments to translate legal obligations into concrete requirements that influence not only which systems are acquired, but how they are designed.

Defense procurement should therefore be understood as more than an administrative mechanism for acquiring military capabilities. It gives governments leverage to influence technological development by communicating legal, ethical, policy, and operational expectations to industry.

Governing Technological Development

This governance function is particularly significant because IHL establishes legal standards without prescribing technical solutions. The principles of distinction, proportionality, and precautions in attack (inter alia) define the legal framework governing the use of military force. Likewise, Article 36 of AP I obliges States to review new weapons, means and methods of warfare before their “study, development, acquisition or adoption” with respect to their compliance with international law. Thus, the legal standard is clearly defined. Yet, how States and industry implement these obligations is naturally not determined.

This question emerges during the design, development and production process within the industry and is answered by AI and software engineers. Selection of training data, software architectures, safety and assurance mechanisms as well as testing (to name just a few) are highly relevant factors which determine an AI-enabled product’s IHL compliance and ethical qualities. Once these decisions have been implemented, they often become difficult and costly to reverse.

Procurement provides States with an opportunity to influence this process. Governments can communicate expectations to industry before fundamental design choices have been made if they include capability requirements, technical specifications, evaluation criteria in the procurement process, and include important obligations for industry in contractual provisions. As scholars observe, procurement represents the practical stage at which commitments to responsible military AI can be translated into operational requirements. Similarly, others argue that governments should communicate their legal expectations early enough to influence system design rather than attempting to address legal concerns once development is already well advanced.

As for practical implementation within the tender process, governments should distinguish between absolutely mandatory legal requirements—meaning that in case of non-compliance a product would be excluded from the tender—and desirable characteristics of responsible military AI, which would improve a product’s chances for tender selection. Compliance with IHL (broken down into verifiable criteria) should form part of the mandatory eligibility requirements for participation in a tender. Beyond these “must-have” requirements, additional characteristics, such as explainability, traceability, robustness or auditability, should be incorporated as weighted evaluation criteria. Manufacturers investing in these capabilities would thereby enjoy a competitive advantage during the award process.

In the long term, procurement can do more than select between competing products, it can shape the product supply itself. By determining which characteristics are rewarded, governments influence where defense companies invest research and development resources and how future military AI systems are designed.

Governing the AI Lifecycle

Through contractual arrangements, the governance function of procurement can be extended throughout the contractual and technical lifecycle of a product. AI-enabled capabilities continue to evolve through software updates, retraining of their AI models, adaptation to operational environments, and new operational data. A meaningful weapons review therefore cannot be limited to a one-time ex ante review of a system. Where a system’s functionality and qualities change beyond a certain threshold, a renewed weapons review must be conducted, reassessing the system in light of its updated capabilities and amended operational use. This may entail renewed testing, evaluation, verification, and potentially require the manufacturer to make further modifications to the changed system.

Procurement provides the framework for implementing these procedural rules already in the supply contract. States must translate their obligation under IHL into enforceable rights against a system’s manufacturer, binding them to comply with repeated weapons reviews, accept change requests, and implement required changes. That way, compliance becomes a continuous performance obligation instead of a one-off requirement.

Incentivizing Responsible Design

Implementing IHL criteria into procurement has a long-term effect beyond single tender processes. Defense companies operate in competitive markets. Sound model training, good training data quality, extensive testing and evaluation, and implementing design choices for responsible military AI require time, expertise, and financial investment. Companies will make these investments only if States consistently recognize their value.

This means that investments in governance may become commercially disadvantageous if procurement focuses primarily on cost, delivery schedules and technical performance. Responsible companies should be rewarded for investing in compliance, robustness, transparent development processes or lifecycle assurance.

Effective procurement should therefore reward responsible innovation. By recognizing these investments in responsible military AI with advantages in procurement processes, governments incentivize industry to invest systematically in responsible development. If several major procuring States adopt comparable procurement criteria, defense manufacturers have strong commercial incentives to integrate these standards across their entire international product portfolios. Procurement could thereby generate an indirect regulatory effect comparable to the “Brussels Effect,” encouraging responsible military AI well beyond the jurisdictions that originally introduced such procurement requirements.

Industry’s Responsibility: From Procurement Requirements to Engineering Practice

Responsible military AI cannot be achieved through procurement alone. How technical, legal, and operational requirements are translated into engineering decisions, organizational processes, and ultimately military capabilities depends upon the companies responsible for developing the products. Responsible military AI therefore relies on industry itself putting such governance in place.

Internal Governance Frameworks

Building such governance within industry requires organizational structures capable of translating procurement requirements into engineering practice. In many defense organizations, legal advisors, compliance officers, software engineers, and ethics teams often operate largely within separate institutional silos. Important decisions are taken in isolation rather than collaboratively, limiting the ability of legal and ethical considerations to influence technical development (and vice versa) from the outset.

Military AI requires a more integrated approach. Legal analysis should influence engineering decisions before technical architectures become fixed. Engineers must understand the legal and ethical implications of design choices. Procurement, compliance, and ethics should support product development throughout the capability lifecycle rather than reviewing completed decisions retrospectively. Design choices should be reviewed at specific, predetermined points in the development process. At these checkpoints—typically aligned with milestones such as requirements freeze, architecture review, and pre-deployment testing—legal, ethics, and engineering representatives jointly assess design choices against defined IHL and procurement criteria, with unresolved conflicts escalated before development proceeds.

I refer to this integrated organizational approach as Integrated Legal-Ethical Engineering Governance (ILEEG). Rather than creating an additional compliance function, the ILEEG framework seeks to connect existing disciplines within a common governance architecture capable of translating procurement requirements into engineering practice.

External governance through procurement and internal governance within industry are therefore complementary. Governments can establish expectations through procurement, which industry must implement through its own integrated governance structures.

Conclusion

Defense procurement can be used as an important lever through which governments shape not only the products they acquire, but more strategically set incentives for steering the direction of ethical technological development.

International and national regulation remains essential: it is the path through which the international community and national lawmakers can set durable, uniform standards that procurement alone cannot establish. Yet governmental procurement remains a valuable complement, an interim measure ahead of binding rules—and on a lasting basis—the level of operational detail general regulation cannot reach. Responsible military AI depends upon organizations capable and willing to implement them into the development process. Defense companies require governance structures that integrate legal, ethical, and technical expertise throughout the development process. External governance through procurement and internal governance within industry are complementary rather than competing approaches.

Governments need not wait for new international agreements before shaping responsible military AI. Every procurement decision already influences how military AI is designed, developed, and maintained. Used strategically, defense procurement is not merely a purchasing function for States but can be a highly effective governance instrument that they already possess.

***

Dr Anke Allenhöfer is Lead Counsel at a defense tech company and a Lieutenant Colonel (Res.) in the German Armed Forces. She serves on the Independent Advisory Board on Legal Reviews of the Responsible AI Institute.

The views expressed are those of the author, and do not necessarily reflect the official position of the United States Military Academy, Department of the Army, or Department of Defense.

Articles of War is a forum for professionals to share opinions and cultivate ideas. Articles of War does not screen articles to fit a particular editorial agenda, nor endorse or advocate material that is published. Authorship does not indicate affiliation with Articles of War, the Lieber Institute, or the United States Military Academy West Point.

Photo credit: Copyleft via Wikimedia Commons