THE LONG VIEW

Bottom Line Up Front: The absence of a binding international framework for military AI creates a high-risk regulatory void, enabling unchecked militarization of dual-use AI technologies and threatening global stability—making EU-led governance efforts urgent and strategically decisive. Threat Identification: Military artificial intelligence (AI) is being rapidly integrated into defense systems worldwide, driven by great-power competition (U.S., China, Russia) and corporate involvement (e.g., Palantir, ClearviewAI, OpenAI). The lack of global governance allows for mission creep, where civilian AI tools are repurposed for lethal autonomous weapons systems (LAWS), surveillance, and algorithmic targeting, undermining international humanitarian law and human oversight [1]. Probability Assessment: The risk of widespread, unregulated military AI deployment is high and already materializing—evident in Ukraine and Gaza by 2024–2026 [2]. Without intervention, fully autonomous AI-driven warfare systems could become operational by 2030. Diplomatic initiatives like REAIM (2023), the U.S. Political Declaration, and the UN resolution on LAWS indicate growing concern but remain nonbinding, offering limited deterrence [3]. Impact Analysis: Unchecked military AI could lead to accidental escalations, adversarial manipulation of targeting systems (e.g., misidentifying civilians), and erosion of meaningful human control (MHC). The dual-use nature of AI complicates export controls and accountability, increasing the risk of proliferation. Ethical and legal responsibilities of private tech firms in conflict zones remain unaddressed, threatening human rights and global security norms [4]. Recommended Actions: The EU should lead in forming a multistakeholder coalition to establish a legally binding treaty on military AI, building on the REAIM process and UN efforts. It must extend its AI Act’s risk-tiered model to dual-use and military applications, prohibit LAWS without human judgment, and strengthen defense-industry guidelines. The EU should also forge minilateral partnerships with like-minded states and tech firms to align governance standards and promote MHC as a global norm [5]. Confidence Matrix: - Threat Existence: High confidence (based on documented battlefield use and corporate policy changes) - Probability of Escalation: Medium-high confidence (due to geopolitical competition and technological momentum) - Impact Severity: High confidence (given potential for catastrophic failures and legal erosion) - Effectiveness of Current Measures: Low confidence (nonbinding initiatives lack enforcement) - Feasibility of EU Leadership: Medium confidence (political will and regulatory capacity exist, but defense remains member-state sensitive) Citations: [1] Csernatoni, R. (2026). *Regulating military AI, a challenge debated in Geneva alongside the G7*. Le Monde.fr. [2] Bergengruen, V. (2023). *The AI War Lab: Ukraine’s Battlefield of the Future*. Time. [3] REAIM Summit (2023); U.S. Political Declaration on Responsible Military Use of AI; UN Resolution on LAWS (2023). [4] OpenAI (2024). *Revised Usage Policies*. [5] Carnegie Europe (2026). *Governing Military AI Amid a Geopolitical Minefield*.