On 14 July 2026, I spoke on behalf of ITU during Panel 8, “The Rights of Indigenous Peoples and Artificial Intelligence,” at the 19th session of the Expert Mechanism on the Rights of Indigenous Peoples (EMRIP), in the Assembly Hall of the Palais des Nations in Geneva. The panel brought together Indigenous representatives from the Arctic, North America, and Eastern Europe/Russian Federation/Central Asia, alongside representatives of Australia and the Global Alliance of National Human Rights Institutions. The full session is archived on UN Web TV.
I built my intervention around three points. Connectivity is still not a fully realized right for Indigenous Peoples. Artificial intelligence can either deepen that exclusion or help repair it, depending on who designs it. And technical standards, the largely invisible rules that shape how technology behaves, are a legitimate and still underused space for Indigenous participation.
On connectivity, I noted that 2.2 billion people worldwide remain unconnected, and that the more than 476 million Indigenous people across some 90 countries, carrying roughly 5,000 distinct cultures and languages, are structurally overrepresented among them. That is not geographic coincidence. It reflects long patterns of dispossession and infrastructure investment that consistently bypassed Indigenous territories. I spoke about ITU’s mandate, renewed through Resolution 46 at last year’s World Telecommunication Development Conference in Baku and grounded in Resolution 184 from our 2022 Plenipotentiary Conference, to support connectivity that is Indigenous led rather than simply delivered. A training programme run with regional partners across nineteen Latin American countries has helped Nasa, Garífuna, Quechua, Wichí, Diaguita-Calchaquí and Guaraní communities, among others, learn to build, run, and maintain their own networks, with close to 120 graduates, over four in ten of them women, going on to train thousands more. The Irundi cooperative network in Argentina, named after the Guaraní word for “four” because it began with four donated routers, is a small but telling example of what community ownership of infrastructure looks like in practice.
On AI, I raised a harder problem. Current systems are trained overwhelmingly on a handful of dominant languages, and estimates presented at this year’s WSIS Forum suggest that up to four billion people, whose mother tongues, cultures, and local knowledge remain underrepresented in training data, risk getting unreliable or distorted results from AI systems now embedded in public services, education, and everyday search. For Indigenous languages, many already endangered after decades of active suppression, that is not a minor accuracy problem. It is a quieter pressure toward erasure. But the same technology, placed under Indigenous control, can work the other way. I pointed to Te Hiku Media in New Zealand, which built an AI tool to collect and preserve oral recordings in te reo Māori and placed the resulting data under a license that keeps governance inside the community and explicitly bars its use for surveillance or discrimination.
Not AI applied to Indigenous Peoples, but AI governed by them, on infrastructure and licensing terms they control.
That is the model I argued should be generalized. On standards, I explained that ITU member states adopted a resolution at the 2024 World Telecommunication Standardization Assembly directing our AI standardization work toward trustworthiness, and that members are now reviewing a simple, voluntary, human centric guideline meant to flag human rights risks early in the standards development process, before systems are built rather than after they cause harm. That is exactly the point where Indigenous data sovereignty, language rights, and free, prior and informed consent need to be built into technical design choices, not added years later through litigation or politics.
I closed by naming three concrete commitments on ITU’s part: continuing to work with Indigenous led community networks, supporting AI for the common good that includes Indigenous communities, and encouraging Indigenous participation in our technical standardization processes. Fielding a question from the floor, I also noted that ITU is actively collaborating with the Coalition for Content Provenance and Authenticity (C2PA) on standardizing media provenance and authenticity to counter disinformation and deepfakes.
Technology did not create the injustices Indigenous Peoples have faced. But built carelessly, it can deepen them. Built in genuine partnership, it can help Indigenous Peoples do what they have always done: preserve their languages, their knowledge, and their self-determination, on their own terms.