As the complexity of climate negotiations grows, artificial intelligence could help negotiators work smarter, not harder — but only if we invest in the skills, safeguards, and systems to use it wisely.
At the recent UNFCCC SB62 meetings in Bonn, negotiators, technologists, and civil society leaders explored an urgent but often overlooked question: How do we improve the process of climate negotiation itself?
The context was clear. In just a few months, COP30 will convene in Belém, Brazil—marking a decade since the adoption of the Paris Agreement. Yet despite progress in ambition and frameworks, many of the systemic challenges facing climate diplomacy remain unchanged. In some cases, they have intensified.
Our side event—The Role of AI in Multilateral Climate Processes—co-hosted by the UNFCCC Secretariat, the incoming Brazilian COP30 Presidency, ComplexChaos, Frontline Associates, and CEMUNE—sought to explore how artificial intelligence could support negotiators in navigating rising complexity, information overload, and persistent coordination gaps.
“Artificial intelligence offers something rare: the potential to level the playing field,” noted Cecilia Njenga, Director at the UNFCCC Secretariat, in her opening remarks. “It can enhance the capacity of underrepresented voices and make negotiations smarter and more inclusive.”
Climate diplomacy has become one of the most complex negotiation environments in the world. At each COP, negotiators are expected to juggle:
These burdens are particularly acute for smaller delegations from the Global South. Many must stretch just two or three delegates across dozens of issues, with little time or bandwidth to coordinate internally, let alone absorb evolving text or track changes in real-time.
This is where AI, done responsibly, can help—not by automating diplomacy, but by supporting the humans at the heart of it.
Brazil’s COP30 Digital Support Project includes an AI assistant, trained on over 2,500 UNFCCC documents and capable of multilingual interaction. It’s already been deployed at COP29 as a pilot and is openly accessible via one the Presidency’s platforms —signaling a commitment to transparency, inclusivity, and innovation.
The project also includes immersive virtual “multiverse rooms” that aim to expand participation for stakeholders who cannot travel or engage in traditional formats. As a model, it exemplifies how digital tools can amplify inclusion—when backed by clear governance and user-centered design.
As a practitioner working to support multilateral negotiations around the world, I see three imperatives that must guide the integration of AI into this space:
Too often, AI tools are developed without meaningful input from those who use them. CEMUNE is partnering with organizations like Harvard, Frontline Associates, and ComplexChaos to co-design training that goes beyond traditional negotiation skills — helping negotiators both use AI tools effectively and shape how these tools are built. This is part of our broader effort to equip negotiators with the full range of skills needed for today’s climate diplomacy.
As Claude Bruderlein of the AI Negotiation Challenge warned:
“The risk isn’t that we use AI. The risk is that we don’t understand how we’re using it—or worse, allow it to quietly rewire the way we negotiate without realizing it.”
Internal misalignment within negotiation groups is a chronic issue that causes delays and, at its worst, marginalises those who most need their voice to be heard. ComplexChaos presented its Alignment Intelligence™ tool, which enables Parties to surface disagreement, misunderstanding, and weak assumptions before they enter formal talks. It doesn’t just summarize input—it highlights divergence, making it easier to focus time and energy on what matters most.
“AI can’t replace the judgment diplomacy requires,” said co-founder Maya Ben Dror. “But it can accelerate the alignment that makes diplomacy work.”
AI tools used in international settings must uphold data protection standards, minimize risk of bias, and offer sovereign deployment options. That includes the ability to run tools offline or on-premise, and to trace how data is being used. We heard strong calls during the session—particularly from Indigenous Peoples and Global South participants—to ensure AI development does not reproduce extractive patterns of knowledge use.
The climate crisis is not just a test of technology or policy—it is a test of our capacity to cooperate at scale. If artificial intelligence can help negotiators listen better, align faster, and design fairer outcomes, then its potential is worth pursuing.
But we must tread carefully. The human dimensions of diplomacy—trust, empathy, judgment—are irreplaceable. AI should not be seen as a shortcut to consensus, but as a scaffold to help us get there with greater clarity and speed.
As we prepare for COP30, let us remember: we’re not only facing a decade of implementation—we’re facing a decade of transformation. And if used wisely, AI can help ensure that transformation leaves no voice behind.
