Ethical AI principles are everywhere in defense policy, yet they rarely translate into actual practice. A sweeping systematic review of 1,085 academic publications has uncovered why: seven interconnected barriers operate like a broken system where fixing one problem simply pushes failure elsewhere. The research reveals that governance and structural obstacles dominate the conversation, but technical, operational, and cultural challenges are equally consequential and far harder to address.

What Are the Seven Barriers Blocking AI Ethics in Defense?

Researchers identified a comprehensive set of obstacles that prevent ethical AI governance from moving from theory to real-world implementation. These barriers don't operate in isolation; they reinforce each other, creating a self-perpetuating cycle of failure. Understanding each one is essential for anyone involved in defense policy, military technology, or AI oversight.

• Governance and Structural Barriers: Lack of clear authority, fragmented decision-making across organizations and nations, and absence of enforceable mechanisms prevent coherent ethical oversight.
• Conceptual Barriers: Disagreement over what "ethical AI" actually means, conflicting definitions of fairness and accountability, and unclear translation of principles into measurable standards.
• Strategic Barriers: Competing military priorities, geopolitical pressures, and institutional incentives that favor speed and capability over ethical compliance.
• Operational Barriers: Difficulty integrating ethical considerations into real-time decision-making, workflow constraints, and pressure to deploy systems faster than governance can keep pace.
• Relational and Cultural Barriers: Organizational silos, lack of trust between stakeholders, and institutional cultures that view ethics as a compliance burden rather than a core capability.
• Technical and Data Barriers: Insufficient transparency in AI systems, poor data quality, and inability to explain algorithmic decisions in high-stakes military contexts.
• Resource, Skills, and Capacity Barriers: Shortage of interdisciplinary expertise, inadequate funding for ethical governance infrastructure, and lack of dedicated personnel.

The prominence of governance and structural barriers in academic literature may reflect their visibility rather than their true importance. Technical, operational, and cultural dynamics remain harder to trace but are equally consequential for whether ethical principles actually work in practice.

Why Does Ethical AI Governance Matter So Much in Defense?

The stakes in defense are extraordinarily high. Ungoverned AI systems in military contexts risk escalating conflicts, violating international humanitarian law, and eroding principles of Just War Theory. There's also the critical issue of insufficient oversight of autonomous systems empowered to apply lethal force. When algorithmic failures occur in defense, the consequences are measured in lives lost, protection of non-combatants compromised, and potential destabilization of international order.

Over the past decade, researchers and policymakers have produced an explosion of ethical principles and frameworks designed to guide AI in defense. Yet despite this proliferation of guidance, actual implementation has proven remarkably resistant to change. The question of why that implementation systematically fails, particularly in the high-risk domains where governance matters most, has received surprisingly little scholarly attention until now.

How to Build Effective AI Ethics Governance in Defense

The research team proposes three strategic interventions that address the systemic nature of these barriers. Rather than piecemeal fixes that simply move problems around, these meta-interventions aim to fundamentally reshape how defense organizations approach ethical AI.

• Establish Authoritative, Interoperable Governance Architectures: Create clear decision-making structures that transcend organizational and national boundaries, enabling coherent oversight across military institutions and allied nations.
• Build Institutional Capacity Through Dedicated Resources: Invest in dedicated funding, personnel, and interdisciplinary expertise specifically tasked with ethical governance rather than treating it as an afterthought to existing operations.
• Integrate Ethical Governance as Core Capability: Embed ethical considerations throughout the entire AI lifecycle, from initial design through deployment and monitoring, rather than bolting compliance onto finished systems.

The research emphasizes that local fixes often merely displace failure elsewhere in the governance system rather than resolving it. If failure is systemic, then piecemeal initiatives will not suffice. Reform must alter the structures and incentives that shape governance choices under geopolitical competition, operational tempo, and institutional path dependence.

What Makes This Research Different From Other AI Ethics Studies?

This systematic review represents the first comprehensive mapping of implementation barriers specifically in the defense domain. Rather than simply cataloguing ethical principles, the research moves beyond abstract values toward understanding the actual institutional, structural, and technical conditions under which those principles can acquire practical force. The analysis examined 1,085 publications and distilled them down to 53 core documents that directly addressed implementation challenges.

The findings have implications far beyond defense. The barriers that obstruct ethical governance of AI in defense are not unique to the military context. Analyzing them illuminates structural challenges common to other high-risk domains, including healthcare, justice, and financial services. Understanding why ethical governance fails at the point of operationalization in defense can help policymakers and institutions in other sectors avoid similar pitfalls.

The research reveals that barriers cluster predominantly at organizational and institutional levels rather than being purely technical problems. This insight has profound policy implications. It suggests that simply developing better algorithms or more transparent AI systems will not solve the governance problem if the underlying institutional structures, incentives, and resource allocation remain unchanged. Meaningful reform requires addressing how organizations are structured, how decisions are made, and how resources are allocated across competing priorities.