Global Rules and Policies for Military AI Control

Introduction

The debate over military AI governance is no longer about a distant future. AI-assisted targeting, autonomous drones, battlefield decision-support systems, and military large language models are already appearing in real conflicts. Governments increasingly see AI as a source of strategic advantage, creating pressure to deploy systems quickly before rivals do. That dynamic raises a central question: can the world gain the defensive, stabilising, and potentially civilisation-protecting benefits of advanced AI without triggering an uncontrolled military arms race?

For advocates of a broader AI-enabled human flourishing, or “AI bloom”, this question matters far beyond defence policy. A future of scientific acceleration, abundance, longer lives, and greater human capability depends on avoiding catastrophic conflict and preserving international cooperation. If advanced AI becomes primarily a tool of military competition, the technologies that might expand humanity’s long-term future could instead intensify instability and mistrust. The challenge for governance is therefore not simply to regulate weapons, but to create conditions in which powerful AI systems are developed under rules, oversight, and norms that reduce incentives for reckless escalation.[SIPRI]sipri.orgImpact of Military Artificial Intelligence on Nuclear…Increasing integration of artificial intelligence (AI) into military system…[SIPRI]sipri.orgArtificial Intelligence, Strategic Stability and Nuclear RiskThis report aims to offer the reader a concrete understanding of how th…

International Treaties and the Search for Global Rules

Why military AI does not fit neatly into existing arms control

Traditional arms control treaties were designed around identifiable physical systems: nuclear warheads, chemical agents, missile launchers, or specific categories of weapons. Military AI is different. It is largely software-based, often dual-use, and frequently embedded within broader systems rather than existing as a standalone weapon.

The same machine-learning techniques that help doctors analyse scans or scientists model proteins can also help militaries process surveillance data, coordinate drone swarms, or identify targets. This makes verification difficult. Unlike nuclear materials, AI models can be copied, modified, and distributed rapidly across borders.[SIPRI]sipri.orgArtificial intelligenceSIPRI's research on AI explores themes such as (a) how AI may find uses in conventional, cyber and nuclear fo…

As a result, many policymakers have shifted from asking whether AI can be controlled through traditional disarmament models toward questions of operational constraints, human oversight requirements, transparency measures, and risk-reduction mechanisms.

The Convention on Certain Conventional Weapons process

The main international forum for discussions on lethal autonomous weapons systems (LAWS) has been the United Nations Convention on Certain Conventional Weapons (CCW). Since 2014, states have met in Geneva to discuss whether new rules should prohibit or regulate systems capable of selecting and attacking targets without direct human control.[Reuters]reuters.comNations meet at UN for 'killer robot' talks as regulation lagsDespite growing reliance on AI in warfare, binding international standards to govern their use remain largely absent. Since 2014, discuss…

A central concept emerging from these discussions is “meaningful human control” or, more recently, “context-appropriate human judgment and control”. The idea is that humans should remain genuinely responsible for decisions involving lethal force rather than serving as nominal overseers of machine-driven processes. Reuters[Springer Link]link.springer.comSpringer LinkLethal autonomous weapon systems (LAWS): meaningful…by S Miller · 2025 · Cited by 5 — This article is concerned with thre…

Despite more than a decade of discussions, progress has been slow. The CCW operates largely by consensus, allowing major military powers to block proposals they consider unacceptable. States disagree not only on regulatory details but also on basic definitions. There is still no universally accepted definition of a lethal autonomous weapon system.[ASIL]asil.orgvolume 29 issue 1Lethal Autonomous Weapons Systems & International Law24 Jan 2025 — On December 2, 2024, the United Nations (UN) General Assembly adop…Published: December 2, 2024 Reuters The result is a familiar arms-control dilemma: countries worried about losing strategic advantages often resist restrictions[reuters.com]reuters.comNations meet at UN for 'killer robot' talks as regulation lagsDespite growing reliance on AI in warfare, binding international standards to govern their use remain largely absent. Since 2014, discuss…, while countries with fewer military AI capabilities tend to favour stronger international rules.

Growing activity in the UN General Assembly

Frustration with slow progress in Geneva has increasingly pushed discussions into the wider United Nations system.

In late 2024, the UN General Assembly adopted a major resolution on lethal autonomous weapons systems with overwhelming support. The resolution encouraged further work on both prohibitions and regulations, reflecting growing international concern that existing frameworks may be insufficient.[ASIL]asil.orgvolume 29 issue 1Lethal Autonomous Weapons Systems & International Law24 Jan 2025 — On December 2, 2024, the United Nations (UN) General Assembly adop…Published: December 2, 2024

Another General Assembly resolution addressed artificial intelligence in the military domain more broadly, affirming that international law applies throughout the life cycle of military AI systems and calling for deeper examination of implications for international peace and security.[Digital Library]digitallibrary.un.orgDigital LibraryLethal autonomous weapons systems:Lethal autonomous weapons systems: resolution / adopted by the General Assembly UN. Ge… [2UNIDIR →] Building a more secure world.

These resolutions are not legally binding. Nevertheless, they help establish diplomatic expectations and create political pressure. Historically, many arms-control initiatives have begun as norms before becoming treaties.

The Core Governance Divide: Ban, Regulate, or Manage?

One of the most important disagreements concerns what exactly should be governed.

Some states, humanitarian organisations, and civil-society groups argue that certain forms of autonomous weapons should be prohibited outright. Their concern is that machines should never independently decide who lives and dies. They argue that such systems create accountability gaps, threaten civilian protection, and undermine fundamental principles of international humanitarian law. UN Regional Info Centre[ASIL]asil.orgvolume 29 issue 1Lethal Autonomous Weapons Systems & International Law24 Jan 2025 — On December 2, 2024, the United Nations (UN) General Assembly adop…Published: December 2, 2024

Others favour regulation rather than prohibition. Their argument is that autonomy exists on a spectrum. Defensive systems such as missile interceptors already operate with substantial automation because human reaction times are often too slow. From this perspective, the goal should be to identify unacceptable applications while allowing beneficial or stabilising uses.[SIPRI]sipri.orgAutonomous Weapon Systems and AI-enabled Decision…This report provides a comparative analysis of autonomous weapon systems and ar…

A third position focuses less on specific technologies and more on outcomes. Under this approach, governance should target risks such as escalation, loss of accountability, accidental engagements, or civilian harm regardless of how much autonomy a system technically possesses.

This debate matters because the future of military AI may involve not only autonomous weapons but also AI-enabled command systems, intelligence analysis tools, cyber capabilities, and increasingly agentic systems capable of planning and coordinating actions over long time horizons. Some researchers argue that existing governance frameworks were developed for earlier generations of autonomy and may struggle to address these newer capabilities.[arXiv]arxiv.orgMeaningful human command: Advance control directives as a method to enable moral and legal responsibility for autonomous weapons sys…

National Military AI Policies Are Becoming a Second Layer of Governance

Because international agreements remain limited, many governments have developed national principles for military AI.

The United States Department of Defense has published policies emphasising responsible AI use, traceability, reliability, governability, and human oversight. Similar efforts have emerged across NATO countries and other technologically advanced states. The emphasis is often on ensuring that commanders remain accountable for operational decisions even when AI systems contribute recommendations or analysis.[The Verge]theverge.comThe Verge AI warfare is already hereInitially viewed as futuristic, lethal autonomous weapon systems (LAWS) are now increasingly present in real-world applications. The tran…

National frameworks can move faster than international treaties, but they face a credibility problem. A government may commit to responsible use while simultaneously investing heavily in military AI capabilities. Rival states may then interpret these investments as evidence that competition is intensifying, regardless of official safeguards.

This is one reason governance experts increasingly focus on interoperability between national approaches. If states can converge around shared definitions, testing standards, audit procedures, and human-control requirements, those norms may gradually become international expectations even without a formal treaty.[SIPRI]sipri.orgLessons from the EU on Confidence-building Measures…This report examines how global governance deliberations might benefit from a…

The challenge is that different countries often have different strategic incentives. Democracies may emphasise transparency and accountability, while authoritarian governments may prioritise speed, secrecy, or operational flexibility. Governance frameworks therefore emerge within broader geopolitical competition rather than outside it.

Confidence-Building Measures May Matter More Than Grand Treaties

Lessons from nuclear arms control

Many observers compare military AI governance to nuclear governance. The comparison is imperfect because AI lacks the clear physical infrastructure of nuclear weapons, but the strategic challenge is similar: preventing fear, uncertainty, and competition from creating dangerous escalation dynamics.

Historically, nuclear stability relied not only on arms-reduction treaties but also on confidence-building measures. These included military hotlines, data exchanges, notification agreements, inspections, and regular diplomatic engagement.

Many researchers argue that military AI may require a comparable toolkit.[SIPRI]sipri.orgLessons from the European Union on Confidence-Building…CONFIDENCE-BUILDING MEASURES IN. DELIBERATIONS ON GLOBAL GOVERNANCE OF. AI IN T…

Practical confidence-building measures under discussion

Rather than attempting immediate global bans, some experts advocate incremental measures that reduce uncertainty and build trust:

• Voluntary transparency regarding military AI doctrines.
• Shared definitions of autonomous functions and human oversight.
• Notification procedures for major military AI deployments or tests.
• Communication channels during crises involving AI-enabled systems.
• Joint technical discussions on accident prevention and escalation risks.
• Shared best practices for testing, validation, and safety assurance.

SIPRI researchers have argued that confidence-building measures may provide a practical path forward because they can generate cooperation even when governments disagree about comprehensive regulation. The European Union’s experience with risk-based AI governance has also been studied as a possible source of lessons for military contexts.[SIPRI]sipri.orgconvenes workshop ai nuclear nexusSIPRI convenes workshop on the AI–nuclear nexus2 Jul 2025 — On 16–17 June, SIPRI held an expert-led workshop centring on AI integration i…

The nuclear-AI nexus

One particularly urgent area involves nuclear command, control, and early-warning systems.

Researchers warn that AI integration could compress decision times during crises, increase pressure for rapid responses, and create new opportunities for misunderstanding. Even AI systems not directly controlling nuclear weapons may influence intelligence assessments, threat detection, or escalation decisions.[SIPRI]sipri.orgco host military ai courseSIPRI to co-host military AI course10 Feb 2026 — June 2025. Lessons from the EU on Confidence-building Measures Around Artificial Intelli…Published: June 2025[SIPRI]sipri.orgOpen source on sipri.org.

Because of these risks, some analysts advocate specific norms limiting AI involvement in nuclear launch decisions. Such norms would resemble earlier agreements that sought to preserve human judgment at moments of potentially civilisation-ending consequence.

For those interested in the long-term future of humanity, this area is especially significant. Advanced AI could eventually contribute to extraordinary scientific and economic progress, but that future depends on avoiding catastrophic conflicts that could derail or destroy the conditions necessary for human flourishing.

Why Arms Race Dynamics Are So Difficult to Control

Military AI creates several pressures that make restraint difficult.

First, software can improve rapidly. A state that delays deployment may fear falling behind competitors.

Second, many AI capabilities are hidden. Governments often cannot accurately assess what rivals possess, encouraging worst-case assumptions.

Third, civilian and military AI ecosystems overlap heavily. Breakthroughs in commercial AI can quickly acquire military relevance.

Fourth, AI may reward speed. Leaders may fear that opponents using AI-enhanced systems will make decisions more quickly, creating incentives to automate further. This can generate a self-reinforcing cycle in which each side interprets the other’s investments as evidence that acceleration is necessary.[SIPRI]sipri.orgmapping military ai industryMapping the military AI industry23 Apr 2026 — The involvement of industry is a key aspiration of arms control initiatives on the responsi…[SIPRI]sipri.orgAI in Chinese, Indian and US Nuclear Postures, Norms and…This survey is intended to serve as a baseline for future engagement with and…

These dynamics resemble classic security dilemmas: actions intended as defensive can appear threatening to rivals. Arms race mitigation therefore requires more than technical safety measures. It requires political mechanisms that reduce fear and improve predictability.

What Success Would Actually Look Like

A realistic vision of military AI governance is unlikely to resemble complete prohibition. AI capabilities are already too widely distributed, too economically important, and too deeply integrated into civilian technology.

Instead, success would probably involve several overlapping layers:

Such arrangements would not eliminate competition. States would still pursue military advantages. The goal would be more modest but still enormously important: preventing competitive pressures from producing uncontrolled escalation, accidental conflict, or the gradual removal of human judgment from decisions with irreversible consequences.[UN Regional Info Centre]unric.orgai in conflict keeping humanity in controlUN Regional Info CentreAI in conflict: keeping humanity in control14 Oct 2025 — The General Assembly's resolution 79/239, adopted in Dece…[SIPRI]sipri.orgPragmatic Approaches to Governance at the ArtificialThis report seeks to advance the discussion by identifying commonalities in risk assessments on the AI–nuclear nexus, examining the…

Why This Matters for Humanity’s Long-Term Future

The strongest case for AI bloom assumes that advanced intelligence can help humanity solve problems that have historically seemed intractable: disease, scarcity, environmental damage, scientific bottlenecks, and perhaps eventually many limits on human flourishing itself.

Yet powerful technologies do not automatically produce beneficial outcomes. They amplify existing institutions, incentives, and geopolitical conditions. If military competition becomes the dominant organising force around advanced AI, humanity could gain extraordinary capabilities while simultaneously increasing risks of conflict, coercion, and catastrophic error.

Global governance frameworks for military AI are therefore not merely constraints on technology. They are attempts to shape the environment in which increasingly powerful intelligence systems develop. The central hope is that cooperation, accountability, and human oversight can grow alongside capability, allowing advanced AI to contribute more to long-term flourishing than to strategic instability.

Whether that balance can be achieved remains uncertain. The history of arms control suggests that governance usually lags behind technology. The history of nuclear weapons, however, also suggests that norms, institutions, and confidence-building measures can emerge even among rivals when the stakes become sufficiently clear. Military AI governance is, in many respects, an attempt to reach that understanding before the technology becomes far more powerful than it is today.[SIPRI]sipri.orgmilitary ai responsible procurementResponsible behaviour in military AI starts with…Oct 16, 2025 — Areas where AI is expected to generate important opportunities in the…[SIPRI]sipri.orgRelated publicationsRomansky, R.: Lessons from the EU on Confidence-building Measures Around Artificial Intelligence in the Military Doma…

Amazon book picks

Further Reading

Books and field guides related to Global Rules and Policies for Military AI Control. Use these as the next step if you want deeper reading beyond the article.

Book

The Age of A. I.

By Henry Kissinger, Eric Schmidt et al.

Discusses AI's impact on international order and strategy.

See on Amazon

Book

Army of None

By Paul Scharre

Covers autonomous weapons governance and military control problems.

See on Amazon

Book

The Coming Wave

By Mustafa Suleyman

Useful for the broader question of AI control and global coordination.

See on Amazon

Book

The Perfect Weapon

By David E. Sanger

Provides adjacent context on arms control problems for dual-use technologies.

See on Amazon

Browse more on Amazon:The Age of A. I.Army of NoneThe Coming Wave

As an Amazon Associate I earn from qualifying purchases.

eBay marketplace picks

Marketplace Samples

Example marketplace items related to this page. Use the search link to explore similar finds on eBay.

Shop location

UsingUSA USA USAUKCanadaAustraliaIreland

Example eBay listing

DJI Drone/ Or model Vinyl WP HQ Decals, Free PnP Easy To Apply

SearcheBay.co.uk: drone model

Browse similar oneBay.co.uk

Example eBay listing

Silent Running Drone Robot Dewey Huey Louie Fully Painted Models Sci-Fi

SearcheBay.co.uk: drone model

Browse similar oneBay.co.uk

Example eBay listing

Xiaomi Fimi X8 SE Quadcopter Drone 2019 Model

SearcheBay.co.uk: drone model

Browse similar oneBay.co.uk

Example eBay listing

Silent Running Drone Robot Dewey Huey Louie Model Kits Sci-Fi

SearcheBay.co.uk: drone model

Browse similar oneBay.co.uk

Browse more oneBay.co.uk

Example items shown for inspiration; availability and pricing can change. Branchoria may earn a commission if you purchase through outbound eBay links.

Endnotes

1. Source: sipri.org
Link:https://www.sipri.org/publications/2025/sipri-insights-peace-and-security/impact-military-artificial-intelligence-nuclear-escalation-risk

2. Source: sipri.org
Link:https://www.sipri.org/publications/2020/policy-reports/artificial-intelligence-strategic-stability-and-nuclear-risk

3. Source: sipri.org
Link:https://www.sipri.org/research/armament-and-disarmament/emerging-military-and-security-technologies/artificial-intelligence

4. Source: reuters.com
Title: Nations meet at UN for ‘killer robot’ talks as regulation lags
Link:https://www.reuters.com/sustainability/society-equity/nations-meet-un-killer-robot-talks-regulation-lags-2025-05-12/

5. Source: reuters.com
Link:https://www.reuters.com/world/progress-rules-lethal-autonomous-weapons-urgently-needed-says-chair-geneva-talks-2026-03-03/

6. Source: link.springer.com
Link:https://link.springer.com/article/10.1007/s10676-025-09874-x

7. Source: arxiv.org
Link:https://arxiv.org/abs/2303.06813

8. Source: asil.org
Title: volume 29 issue 1
Link:https://asil.org/insights/volume-29-issue-1/

Published: December 2, 2024

9. Source: unidir.org
Title: UN General Assembly A RES 79 239 EN
Link:https://unidir.org/wp-content/uploads/2025/03/UN_General_Assembly_A_RES_79_239-EN.pdf

10. Source: sipri.org
Link:https://www.sipri.org/publications/2025/other-publications/autonomous-weapon-systems-and-ai-enabled-decision-support-systems-military-targeting-comparison-and

11. Source: arxiv.org
Title: arXiv The Controllability Trap: A Governance Framework for Military AI Agents
Link:https://arxiv.org/abs/2603.03515

Published: March 3, 2026

12. Source: arxiv.org
Link:https://arxiv.org/abs/2604.06300

13. Source: sipri.org
Link:https://www.sipri.org/publications/2025/eu-non-proliferation-and-disarmament-papers/lessons-eu-confidence-building-measures-around-artificial-intelligence-military-domain

14. Source: sipri.org
Link:https://www.sipri.org/sites/default/files/2025-05/eunpdc_no_97.pdf

15. Source: sipri.org
Title: convenes workshop ai nuclear nexus
Link:https://www.sipri.org/news/2025/sipri-convenes-workshop-ai-nuclear-nexus

16. Source: sipri.org
Title: co host military ai course
Link:https://www.sipri.org/news/2026/sipri-co-host-military-ai-course

Published: June 2025

17. Source: sipri.org
Link:https://www.sipri.org/

18. Source: sipri.org
Title: mapping military ai industry
Link:https://www.sipri.org/commentary/topical-backgrounder/2026/mapping-military-ai-industry

19. Source: sipri.org
Link:https://www.sipri.org/publications/2026/other-publications/ai-chinese-indian-and-us-nuclear-postures-norms-and-systems

20. Source: sipri.org
Title: Pragmatic Approaches to Governance at the Artificial
Link:https://www.sipri.org/publications/2025/other-publications/pragmatic-approaches-governance-artificial-intelligence-nuclear-nexus

21. Source: sipri.org
Title: military ai responsible procurement
Link:https://www.sipri.org/commentary/essay/2025/military-ai-responsible-procurement

22. Source: sipri.org
Link:https://www.sipri.org/research/armament-and-disarmament/emerging-military-and-security-technologies/artificial-intelligence/recent-pubs

23. Source: sipri.org
Link:https://www.sipri.org/publications/2025/sipri-insights-peace-and-security/advancing-governance-nexus-artificial-intelligence-and-nuclear-weapons

24. Source: sipri.org
Link:https://www.sipri.org/sites/default/files/2026-02/0226_milai_procurement.pdf

25. Source: unidir.org
Link:https://unidir.org/files/2022-12/Confidence-Building_Final.pdf

26. Source: eucyberdirect.eu
Link:https://eucyberdirect.eu/events/navigating-the-international-and-european-governance-of-military-ai-from-confidence-building-measures-to-risk-based-[guardrails

27. Source: reuters.com
Title: Russia says US did not grant visa for official to attend UN meeting
Link:https://www.reuters.com/world/russia-says-us-did-not-grant-visa-vice-minister-attend-un-2026-05-26/

28. Source: youtube.com
Link:https://www.youtube.com/watch?v=wh6gAXaGGNs

29. Source: digitallibrary.un.org
Link:https://digitallibrary.un.org/record/4071100

30. Source: digitallibrary.un.org
Link:https://digitallibrary.un.org/record/4071348?ln=en

31. Source: unric.org
Title: ai in conflict keeping humanity in control
Link:https://unric.org/en/ai-in-conflict-keeping-humanity-in-control/

32. Source: unric.org
Title: un addresses ai and the dangers of lethal autonomous weapons systems
Link:https://unric.org/en/un-addresses-ai-and-the-dangers-of-lethal-autonomous-weapons-systems/

33. Source: theverge.com
Title: The Verge AI warfare is already here
Link:https://www.theverge.com/ai-artificial-intelligence/937028/military-ai-warfare-red-lines

34. Source: un.org
Link:https://www.un.org/en/

35. Source: Wikipedia
Title: United Nations
Link:https://en.wikipedia.org/wiki/United_Nations

Additional References

36. Source: reachingcriticalwill.org
Link:https://reachingcriticalwill.org/resources/publications-and-research/publications/17337-wilpf-submission-to-the-un-secretary-general-s-report-on-artificial-intelligence-in-the-military-domain

37. Source: table.media
Link:https://table.media/en/security/feature/ai-guidelines-how-the-federal-ministry-of-defense-plans-to-ensure-human-oversight-in-autonomous-systems

38. Source: lemonde.fr
Link:https://www.lemonde.fr/en/opinion/article/2026/03/29/the-use-of-lethal-autonomous-weapons-controlled-by-ai-goes-against-the-principles-of-a-just-war_6751919_23.html

39. Source: publications.parliament.uk
Link:https://publications.parliament.uk/pa/ld5804/ldselect/ldaiwe/16/1608.htm

40. Source: unipd-centrodirittiumani.it
Link:https://unipd-centrodirittiumani.it/en/news/united-nations-resolution-79l77-adopted-by-the-general-assembly-on-lethal-autonomous-weapons-systems

Published: December 2, 2024

41. Source: lieber.westpoint.edu
Title: future warfare national positions governance lethal autonomous weapons systems
Link:https://lieber.westpoint.edu/future-warfare-national-positions-governance-lethal-autonomous-weapons-systems/

42. Source: icrc.org
Title: we cannot let AI be deployed on battlefield without oversight and regulation
Link:https://www.icrc.org/en/statement/we-cannot-let-AI-be-deployed-on-battlefield-without-oversight-and-regulation

43. Source: ukstopkillerrobots.org.uk
Title: reports of ai enabled targeting in iran bring real world impacts to the un
Link:https://ukstopkillerrobots.org.uk/2026/03/06/reports-of-ai-enabled-targeting-in-iran-bring-real-world-impacts-to-the-un/

44. Source: stopkillerrobots.org
Title: reports of ai enabled targeting in iran bring real world impacts to the un
Link:https://www.stopkillerrobots.org/news/reports-of-ai-enabled-targeting-in-iran-bring-real-world-impacts-to-the-un/

45. Source: trendsresearch.org
Title: governing lethal autonomous weapons the future of warfare and military ai
Link:https://trendsresearch.org/insight/governing-lethal-autonomous-weapons-the-future-of-warfare-and-military-ai/?srsltid=AfmBOopPE_eiCRFYa1KmUAhz-spvEhxFZIyYWR4AfoJwWBMfHB2Q0kox

Published: December 2024