Governance After Babel: AI and the Crisis of Multilateralism

In the story of the Tower of Babel, humanity’s power lies in perfect alignment: one people, one language, a shared purpose, and the capacity to act toward a common vision - until that shared language is fractured as punishment for hubris. For much of the UN’s existence, multilateralism rested on a comparable assumption: that shared rules, norms, and institutions could serve as the unifying glue that disciplines power and renders collective governance both legitimate and effective. Today, that assumption no longer holds, and we seem to have entered an “after Babel” era of fragmented multilateralism.

The United Nations faces one of the most serious institutional crises in its history: a persistent liquidity crisis, a paralyzed Security Council, intensifying great-power rivalry, and declining confidence in multilateral problem-solving. Strategic competition, rapid technological change, and diverging political visions have not just strained cooperation - they have eroded the shared assumptions that once made cooperation possible.

In this new landscape, the UN is seeking to position itself as a convenor for global governance of Artificial Intelligence, arguably the most transformative technology of the century. The organization is attempting to coordinate action around a technology that is reshaping power faster than multilateral processes can adapt.

The launch of the UN’s Independent International Scientific Panel on Artificial Intelligence and the Global Dialogue on AI Governance reflects recognition that AI’s implications for peace and security, development, human rights, and economic power cannot be addressed through national frameworks alone. But it also reveals the core challenge: the mechanisms designed to govern AI depend on consensus, voluntary compliance, and at least partial normative alignment - precisely the conditions now under pressure.

This article does not ask whether that tension is sustainable (history suggests it is not), but how the UN can navigate it strategically and what governance after Babel can realistically look like.

Fragmented Governance and Divergent Objectives

The global AI governance landscape reflects fundamentally different strategic goals. The European Union’s AI Act establishes a risk-based regulatory framework rooted in product-safety law and fundamental-rights protections, prioritizing accountability even where this may constrain speed of deployment. China has adopted a state-centric approach focused on rapid industrial integration, self-reliance in chips, and strategic application across industrial and governmental sectors. The United States relies on a decentralized, sector-based model, and its America’s AI Action Plan frames AI development in explicitly strategic and competitive terms.

Each of these models is internally coherent, but none is designed to converge into a single global regulator, and diplomatic effort is unlikely to erase these differences. The objective is therefore not convergence, but manageability. Harmonization may be unlikely, but coordination remains possible. For many states outside the major AI powers, participation in shaping rules is limited. They are more often exposed to standards, technologies, and norms developed elsewhere. Hence, AI governance is already pluralistic. The question is whether the UN can play a role in that pluralism that is substantive rather than purely symbolic.

Technology as the New Strategic Terrain

The broader security landscape reinforces this shift. Whatever remained of assumptions about a stable post–Cold War order has largely dissolved. Technology now sits squarely at the center of strategic competition, and control over compute, cloud infrastructure, semiconductor supply chains, and foundational platforms is increasingly treated as geopolitical leverage.

Across advanced economies and developing states alike, reliance on externally controlled digital infrastructure has emerged as a structural vulnerability. In this environment, AI governance frameworks determine who sets standards, who carries risk, and who must eventually adapt to systems they did not design and that may not align with their national interests. This shift blurs the distinction between “technology policy” and “security policy,” and it reframes global AI discussions as contests over influence more so than cooperation. For multilateral institutions built on voluntary compliance and shared interest, this context is deeply destabilizing.

How AI Governance Exposes the UN’s Institutional Constraints

AI governance does not create the UN’s institutional constraints but magnifies them.

Speed. Multilateral processes operate on consensus-driven timelines, sometimes measured in years. The Global Dialogue on AI Governance (A/RES/79/325), for instance, will convene annually to exchange best practices and promote cooperation. AI development cycles operate in months and are accelerating. This does not invalidate UN processes. But it means the organization cannot credibly position itself at the technological frontier in real time. Overstating that capacity risks undermining trust.

Concentration versus universality. Advanced AI capabilities are concentrated among a small number of companies and countries. The UN’s strength lies in universal representation. Most Member States deliberate on frameworks shaped by technologies they neither develop nor control. This is not a gap goodwill alone can close. It reflects the structure of the AI economy.

Acceleration of labor disruption. Beyond regulatory coordination, AI raises deeper socioeconomic risks. If current capability trends translate into large-scale automation of cognitive work, disruption could unfold faster than institutions can adapt. Many industry leaders argue that major segments of knowledge work could be transformed within years rather than decades. Whether these projections prove exact is less important than the risk they signal: speed. This acceleration will not be evenly distributed. States with strong AI ecosystems and fiscal buffers may capture productivity gains while cushioning labor displacement. Economies with limited digital infrastructure may experience disruption without equivalent capacity to respond. The result could be widening inequalities between countries and internal instability within them. AI is therefore not only a productivity tool. It is a volatility multiplier. And its governance is not only a question about oversight. It is also about narrowing the gap between disruption and policy response.

Dual-use complexity. AI systems are inherently dual-use. Military applications are debated in forums such as the CCW process on lethal autonomous weapons systems and in Security Council discussions, largely separate from civilian AI initiatives. Yet general-purpose systems blur these boundaries. Dual-use governance is not new: the Treaty on the Non-Proliferation of Nuclear Weapons (NPT), Chemical Weapons Convention (CWC), and Biological Weapons Convention (BWC) regimes provide historical reference points. AI compounds the challenge because identical systems underpin both civilian and military applications. There is currently no formal institutional mechanism ensuring that civilian AI risk discussions systematically inform security deliberations — or vice versa. A structured liaison function could begin to address that gap without merging mandates.

Financial and institutional fragility. AI governance mechanisms within the UN rely heavily on voluntary contributions at a time of chronic liquidity strain. When core coordination functions depend on voluntary funding, independence cannot simply be assumed. Voluntary structures may concentrate influence among the states most able and willing to contribute — often those with the largest AI industries. Resolution A/RES/79/325 emphasizes independence and balanced participation. Sustainable, insulated funding remains unresolved. This does not predetermine outcomes, but it narrows room for maneuver.

Where the UN Retains Strategic Value

1. Build System-Wide Institutional Memory. AI is already used across the UN system: in humanitarian logistics, refugee protection and flow prediction, disease surveillance, conflict early warning and more. Yet there is no cross-agency mechanism for documenting failures, biased outputs, or unintended consequences. A protected internal AI incident registry – possibly modeled loosely on aviation safety reporting – would enable structured learning without creating public liability exposure. Hosted by the Scientific Panel, it would generate an evidence base no other institution could assemble.

2. Produce an Annual AI Interoperability Map. Rather than another broad “state of AI” report, the Scientific Panel could map how major regulatory frameworks interact in practice: where compliance overlaps, where conflicts emerge, and where gaps leave developing countries exposed. This would make fragmentation navigable without pretending it can be eliminated.

3. Establish a Civilian–Security Liaison Function. A light-touch liaison between civilian AI processes and security-focused discussions would allow structured information exchange without collapsing mandates. General-purpose AI systems do not respect procedural silos. A liaison mechanism acknowledges that reality while remaining politically feasible.

4. Prioritize Technical Coordination Infrastructure. Some of the UN’s most durable coordination successes – aviation safety standards, telecommunications spectrum allocation – succeeded because they focused on interoperability rather than political bargaining. Shared incident taxonomies, documentation templates, and voluntary conformity tools may seem modest. But incremental coordination often proves more durable than high-level declarations. Common reporting standards would also strengthen security-related mechanisms and reporting systems by ensuring data is interoperable and usable across the system.

5. Reduce Asymmetry in AI Adaptation Capacity. The UN cannot slow technological acceleration. But it can help narrow adaptation gaps and prevent widening disparities between advanced and developing economies. Structured partnerships with major AI providers—focused not only on model access but also on compute allocation, multilingual training data, and technical capacity development—could strengthen implementation capacity in developing contexts. In other sectors, such as health, pooled procurement has shown that coordinated demand can improve access terms beyond what individual countries could secure alone. A carefully scoped pilot in AI capacity partnerships could test whether similar mechanisms are feasible. Most of these steps do not require new mandates or dramatic institutional expansion. They require discipline: prioritizing technical coordination over political theater, speed over comprehensiveness, and institutional learning over rhetorical ambition. They also require procedural transparency, particularly around funding sources and multi-stakeholder involvement, to ensure that conflicts of interest are visible and manageable rather than embedded and ignored.

Conclusion: Governing After Babel

The lesson of Babel is not that collective action is inherently doomed to fail. In Babel, language fragmentation was punishment for hubris. We are already operating in an “after Babel” world of multilateral fragmentation. The story therefore serves as a cautionary tale for how the UN positions itself in this new landscape: Coordination and governance fail when they attempt to transcend their own limits. In a fragmented geopolitical “after Babel” world, AI governance must be calibrated to what is feasible — not to what would be ideal.

The success of the United Nations in the era of AI will not be measured by whether it can “govern” Artificial Intelligence – it cannot. It will be measured by whether the organization can contribute substantively and credibly to preventing the already existing fragmentation of the rules-based system from hardening into systemic incoherence – whether it can create just enough shared structure to keep coordination possible in a world defined by competition.

Harmonization may be unlikely, but coordination is still possible. In an era where technological power is concentrating and accelerating, that distinction may determine whether multilateralism adapts or becomes sidelined.

What are the thoughts that keep YOU up at night? Share your questions in the comments - how will AI reshape our international order? Your ideas will help set the agenda for upcoming articles.

If you found this valuable, share it with your network. I write more on emerging technologies, geopolitics, and international peace and security on my Substack: Thoughts That Keep You Up at Night