UN Open Source Week 2025. Part I: Hackathons, Labor Shifts, and AI in the Public Interest
Last week, I had the chance to participate in UN Open Source Week 2025 at United Nations Headquarters in New York: a global gathering of technologists, diplomats, open source developers, and policy advocates building the next generation of Digital Public Infrastructure (DPI). Held from Monday, June 16 through Friday, June 20, the week featured hackathons, side events, and multi-track sessions focused on making digital systems more inclusive, ethical, and worthy of public trust.
Back in our previous post, we previewed what to expect: from governance deep dives to global case studies. What unfolded across five days went beyond expectations, including a panel on safeguards and inclusion in biometric systems, and my role as an (unofficial) chronicler of the week.
For today, I had planned to publish a recap of UN Open Source Week in one sweeping post. But as I reviewed each day’s sessions and transcripts, I realized there was no way to do it justice in a single piece. Too much was said, and much of it is worth preserving, quoting, and learning from.
So instead, we’re launching a five-part series. Each entry will offer a detailed look at what happened on a specific day (or group of days), with reflections on why it matters and what it means for digital governance, open source, and the future of public tech.
Here’s the full release schedule:
Part I – Days 1 & 2, including the LinkedIn side event – Published today
Part II – Wednesday (OSPOs for Good) – Coming Friday, June 28
Part III – Thursday (DPI Day) – Monday, July 1
Part IV – Friday’s community-driven sessions (PwC, RISE, more) – Friday, July 5
Part V – Closing reflections on the week – Monday, July 8
Let’s dive in.
Monday: Hackathons and the Opening Mood
I stopped by the UN on Monday morning to pick up my access pass for the week. There was a long line by the 46th Street entrance that diverted into three or four tables where the passes were being held by first names. Once I held mine in my hand, I passed through security and navigated into the square tower.
Approaching the ECOSOC Chamber, curiosity pulled me into the opening space; a room where the first signals were already being tested as teams spread across rows of tables inside the UNDP Accelerator Lab, dived into code.
The official programming began not with a keynote, but with code. Inside the chamber, contributors were already deep into three concurrent working sessions:
Ahead of the Storm, a hackathon focused on open geospatial tools for climate emergency response,
an Edit-A-Thon on the UN’s institutional memory and Wikipedia presence, and
a Maintain-A-Thon exploring the realities of long-term open source stewardship.
I didn’t stay long, just long enough to get a sense of the pace of people debugging, reviewing, and translating. You could tell this wasn’t just a warm-up.
It felt more like the scaffolding being laid before the stage had even gone up.
TWC Insight: Digital infrastructure doesn’t begin with law or policy. It begins with maintenance, collaboration, and code someone decides is worth improving.
Tuesday: AI, Work, and the Politics of Platform Power
The first day with a side event brought us to LinkedIn’s Midtown offices, where the focus turned to AI, labor, and the infrastructures behind employment itself.
Before its start, I met fellow atendees Mario Cistaro and Michael Anthony Schuler, and we readily began discussing tecnology in general as well as AI before the session (co-hosted with ODET), titled “AI and the Future of Work: The ICT Sector in Transition”, started. So, off to a great start!
But the conversation moved quickly from us to the panelists and from theory into practice. Indeed, Zach Shalvarjian, LinkedIn’s NY Site Lead, traced the company’s evolution from early machine learning to its newer AI tools. “These tools don’t work in isolation,” he emphasized. “We think this technology works best when it’s wrapped around the power of connectedness and human wisdom.”
That note on collective intelligence was echoed by Omkhar Arasaratnam, Distinguished Engineer at LinkedIn, who spoke of his own history with open source. “Open source is a great democratizer,” he said, challenging the room to consider how AI systems might deliver equity (not just efficiency) across borders.
There were two breakout sessions. In the first, “Open Source Skills in the AI Era — Perspectives from the European Open Source Academy” Nicholas Gates moderated panelists Amandine Le Pape (Element; The Matrix.org Foundation; European Open Source Academy); David Cuartielles (Arduino; Malmö University; European Open Source Academy) and Sachiko Muto (RISE Research Institutes of Sweden; OpenForum Europe).
I joined the second: “Advancing Open Source AI for the Public Good,” co-led by Amreen Taneja (Digital Public Goods Alliance) and Stefano Maffulli (Open Source Initiative). Their presentations laid out evolving standards for defining open-source AI: not just in code, but in training data, documentation, and governance. Stefano warned of the “broken social contract” behind current AI development, where public data is captured for private ends. Amreen walked us through the new requirements for qualifying an AI system as a Digital Public Good, emphasizing the need for open datasets and “do no harm” design principles. The standards may be ambitious, she said, “but we believe this is the right call.”
The breakout activity closed with the challenge of exploring incentives for the development and use of AI systems through policy. I joined Sandy Ramírez Pérez, Tashi Gurung, José Díaz, Luz Felix, and Sovieski Naut Fernández (a sharp group largely from the Dominican Republic) and we wrestled with what it would take to turn good intent into good regulation. A genuinely very simple but most complex question, grounded in reality, that I found very thought-provoking; and while there are no easy answers, the conversation itself made room for better ones.
Later in the afternoon, we regrouped for a panel moderated by Lucia Velasco (ODET), featuring Sarah Steinberg (LinkedIn), Craig Ramlal (University of the West Indies), Armando José Manzueta Peña (Vice Minister for Public Innovation, Dominican Republic), and Mehdi Snene (UN ODET). Steinberg spoke candidly about LinkedIn’s labor data and the global skills gap AI is already widening. “Even if you’re not changing jobs,” she said, “your job is changing on you.” Mehdi, meanwhile, laid out a nuanced model of how AI’s sectoral impact on jobs could be assessed beyond hype, emphasizing that “AI for AI” misses the mark: it’s the sectoral shifts that matter.
The panel closed with a call for policies that balance innovation with equity (and for educational models that are more about skills than diplomas). If there was a core thread running through the day, it was this: we can’t let automation define the terms of inclusion.
Before lunch, I had the chance to ask Sarah Steinberg a on how LinkedIn’s labor market data might reflect the rising demand for cybersecurity roles in the AI era. While we only had a brief exchange, we agreed to follow-up; there might be a post here from that future interaction, we’ll keep you posted.
And finally lunch. I found Daniel Shown in the line and we sat with Katie Steen-James and Gabriel Toscano, and were later joined by panelist Amreen Taneja. We continued a thread we’d started earlier: whether “public interest” can be meaningfully preserved on platforms shaped by private logic; the advantages and disadvantages of AI in our daily lives and work. We were so invested in learning about our respective opinions that perhaps we were the last guests to leave the floor.
While there were no marble halls to frame our conversations, the questions were sharp and the discussions carried insight and enthusiasm, that the distance between open source and global labor policy felt surprisingly small.
TWC Insight
Trust in digital systems begins with people. And sometimes, the best technical question is: who gets to participate?
Takeaway
Open source isn't just a development model. It’s an invitation to rethink the terms of inclusion; and to build public infrastructure with public values at the core.
FAQs
1. What is UN Open Source Week?
A weeklong event hosted at UN Headquarters (June 16–20, 2025) focused on open source, public tech, and digital governance. It featured panels, workshops, hackathons, and side events with international speakers and technical experts.
2. Why is open source part of the UN agenda?
Because open code creates more transparent, inclusive systems. From climate monitoring to cross-border customs, open source enables trust, adaptability, and scale. Especially across jurisdictions.
3. What is DPI (Digital Public Infrastructure)?
DPI refers to foundational digital systems — such as digital identity, payments, and data exchanges — that support public service delivery and societal coordination. When built openly, they can serve as digital public goods.
4. What is a Digital Public Good (DPG)?
A DPG is a tech product (like code, content, or data) that’s open source, privacy-respecting, and designed to solve a problem in the public interest. The Digital Public Goods Alliance maintains a global registry.
5. What did the LinkedIn side event focus on?
Titled “AI and the Future of Work,” the event examined how AI is transforming labor markets, job skills, and public policy. It included panels and breakout groups co-hosted with the UN Office of Digital Engagement and Technologies (ODET).
6. What were the two breakout sessions at LinkedIn’s side event?
One focused on open source skills in the AI era, featuring European academics and technologists. The other explored evolving standards for what counts as “open source AI,” including licensing, documentation, and data access.
7. What counts as an “open source AI” system?
It must go beyond code. Open-source AI includes open training datasets, transparent documentation, community governance, and safety practices.
8. How is AI labor being tracked or evaluated?
LinkedIn’s labor market data is one lens. Speakers highlighted how AI is reshaping jobs sector-by-sector, not just replacing them. Skills development, policy incentives, and access equity were key themes.
9. Why does it matter whether AI tools are open?
Because many AI systems are trained on public data. Without openness, those tools often benefit private platforms while eroding public control and accountability.
10. What does “AI for AI” miss, as Mehdi Snene warned?
He argued that evaluating AI only on its own terms misses the real-world impact on labor, education, and inclusion. The real story is how AI transforms existing systems, not just how advanced it gets.
Wrap-up
We’ll be back on Friday with Part II, where we dive into OSPOs for Good: a full-day track on how Open Source Program Offices (OSPOs) are evolving across governments, nonprofits, and tech companies.
If you attended UN Open Source Week, we’d love to hear your take. And if you missed it, we hope this series brings the sessions to life.
Until then, keep an eye on the commits and the conversations.
— Jorge
