Voices is a new mini-series from Humanitarian AI Today. In daily five-minute flashpods we pass the mic to humanitarian experts and technology pioneers, to hear about new projects, events, and perspectives on topics of importance to the humanitarian community. In this flashpod, Cole Leng, an AI Researcher at Harvard and former Project Manager with Nexa AI, sits down with Humanitarian AI Today producer Brent Phillips to discuss the state-of-the-art and future trajectory of on-device machine learning. The discussion provides researchers and staff from humanitarian organizations with important insights into where the state-of-the-art in on-device machine learning stands today and where the cutting-edge is heading. Cole examines the critical trade-offs between on-device and cloud models, analyzing their respective workflows, performance limitations, and implementation considerations to help listeners evaluate whether on-device ML applications are suitable for their specific needs. He also offers insight into choosing LLMs, the impact of new specialized hardware on performance and capability, and how current advances in ML are shaping the next generation of applications. This episode highlights a core goal of the Humanitarian AI Today podcast: fostering dialogue between technology researchers and humanitarian practitioners. As AI and ML capabilities scale rapidly, this cross-sector engagement is crucial for mapping new technical advances to the unique operational, privacy, and resource constraints of the humanitarian field. Substack notes: https://humanitarianaitoday.substack.com/p/cole-leng-on-the-state-and-future

Voices is a new mini-series from Humanitarian AI Today. In daily five-minute flashpods we pass the mic to humanitarian experts and technology pioneers, to hear about new projects, events, and perspectives on topics of importance to the humanitarian community. In this flashpod, Neha Bajwa, Vice President of Product Marketing at Neo4j, joins Humanitarian AI Today producer Brent Phillips to discuss Neo4j’s major plans to support agentic AI systems that will directly benefit nonprofits and humanitarian organizations. Neo4j is the world’s leading graph intelligence platform for AI systems, and the company’s two new tools, Neo4j Aura Agent and the Model Context Protocol Server for Neo4j, will address critical development obstacles and help organizations rapidly build, test, and deploy AI agents. They also discuss Neo4j’s Graphs for Good program and Neo4j’s upcoming “NODES” developer conference, which is the biggest graph community gathering dedicated to applications, data intelligence, knowledge graphs, and AI. Substack notes: https://humanitarianaitoday.substack.com/p/neha-bajwa-from-neo4j-unveils-new

Voices is a new mini-series from Humanitarian AI Today. In daily five-minute flashpods we pass the mic to humanitarian experts and technology pioneers, to hear about new projects, events, and perspectives on topics of importance to the humanitarian community. In this flashpod, Lindsey Moore, Founder and CEO of DevelopMetrics, joins Humanitarian AI Today producer Brent Phillips to talk about agentic AI and bold new funding initiatives like Humanity AI. According to Humanity AI’s website, the coalition is co-chaired by Omidyar Network and the John D. and Catherine T. MacArthur Foundation. Its founding members include the Doris Duke Foundation, Ford Foundation, Lumina Foundation, Kapor Foundation, Mellon Foundation, Mozilla Foundation, Siegel Family Endowment, and the David and Lucile Packard Foundation. Over five-years, the coalition plans to dedicate $500M towards making sure people and communities beyond Silicon Valley have a stake in the future of artificial intelligence establishing an AI future where people and communities can flourish. Lindsey Moore offers a grounded perspective on the current state of AI in the humanitarian sector. She observes that despite industry buzz, most organizations are not yet experimenting with agentic AI. Instead, they are focusing on more foundational challenges, such as organizing their data and building domain-specific large language models that can grasp the unique context and terminology of their work. Discussing the Humanity AI initiative, Lindsey and Brent express hope that such bold new funding can, with an informed understanding of the sector, help offset the destructive impact of recent aid funding cuts. These cuts disrupted foundational AI capacity-building that was being carried out by established organizations with deep domain experience, destroying numerous projects and dismantling teams behind them. They make a compelling case for funders to reinvest in these organizations and their important work to prevent hard-won gains from being lost and to scale their forward momentum. The conversation serves as a call to action, emphasizing that for initiatives like Humanity AI to be truly transformative, they must go beyond traditional philanthropy. This means proactively identifying and engaging with the humanitarian community's own AI leaders, reforming conventional cyclic grant solicitation and grant-making processes that too often overlook real sector innovators and builders, and boldly forming new partnerships that make them accessible and open to input, rather than insulated behind institutional firewalls. Substack notes: https://humanitarianaitoday.substack.com/p/lindsey-moore-from-developmetrics

Voices is a new mini-series from Humanitarian AI Today. In daily five-minute flashpods we pass the mic to humanitarian experts and technology pioneers, to hear about new projects, events, and perspectives on topics of importance to the humanitarian community. In this flashpod, Chelsea McMurray, founder of the AI security startup Dorcha, joins Humanitarian AI Today producer Brent Phillips to discuss international human rights law, AI security, and the threat landscape facing humanitarian actors. They begin with Chelsea’s background in human rights law and the recent disregard for international norms that should underpin ethical AI governance. The casual conversation then pivots to AI security and the specific threats humanitarian organizations face. Chelsea explains how her startup addresses data privacy vulnerabilities and prompt injection attacks, by giving users greater control over their personal information. Protecting such sensitive data is especially critical in the humanitarian sector, where information leaks can endanger field staff and the vulnerable populations they serve. Substack notes: https://humanitarianaitoday.substack.com/p/chelsea-mcmurray-on-ai-security-and

Voices is a new mini-series from Humanitarian AI Today. In daily five-minute flashpods we pass the mic to humanitarian experts and technology pioneers, to hear about new projects, events, and perspectives on critical topics. In this flashpod, Annie Brown, a Data Scientist with Humane Intelligence, talks about her team’s Bias Bounty program and how to get involved in an interview with Brent Phillips, Producer of Humanitarian AI Today. They discuss Humane Intelligence's work focusing on collaboratively designing and running rigorous evaluations that make AI systems more accountable, responsible, and fair, their bias bounty program and the strategy behind it as well as touch on how volunteers can get involved and launch their research. Substack notes: https://humanitarianaitoday.substack.com/p/annie-brown-from-humane-intelligence

Voices is a new mini-series from Humanitarian AI Today. In daily five-minute flashpods we pass the mic to innovators, researchers and practitioners on the humanitarian front lines, delivering real-time news on how they are building, testing and collaborating on uses of artificial intelligence. In this flashpod, Petya Kangalova, Technology Partnerships and Engagement Lead with Humanitarian Open Street Map joins Brent Phillips, Humanitarian AI Today podcast Producer, to discuss Humanitarian Open Street Map’s Technology and Innovation Working Group, its monthly working group open discussion sessions and how people can tune-in. Substack Notes: https://humanitarianaitoday.substack.com/p/petya-kangalova-introduces-humanitarian

Voices is a new mini-series from Humanitarian AI Today. In daily five-minute flashpods we pass the mic to innovators, researchers and practitioners on the humanitarian front lines, delivering real-time news on how they are building, testing and collaborating on uses of artificial intelligence. In this flashpod, Erica Gralla, an Associate Professor at George Washington University , shares news about a new project seeking to understand the impact of recent aid funding cuts on the humanitarian and development system. The study specifically looks at how major funding cuts to U.S. aid programs in 2025 are affecting the sector. She calls on listeners who work in aid or development to participate in the study by taking a "global pulse survey". This project brings together three professors from the fields of engineering, policy, and international relations. Their goal is to understand how the funding cuts are affecting relationships, coordination, information sharing, and supply chains across organizations. By tracking how the aid ecosystem is adapting, the team hopes to capture lessons from this challenging period and help the humanitarian community chart a path forward. Erica and Humanitarian AI Today producer, Brent Phillips, discuss the survey in detail, how people can get involved, what the research team expects to learn, and the project's next steps. Survey Link: http://go.gwu.edu/AidTrack Substack Notes: https://humanitarianaitoday.substack.com/p/erica-gralla-from-george-washington

Voices is a new mini-series from Humanitarian AI Today. In daily five-minute flashpods we pass the mic to innovators, researchers and practitioners on the humanitarian front lines, delivering real-time news on how they are building, testing and collaborating on uses of artificial intelligence. In this podcast episode, Bill Greer, Co-founder of Common Space joins Senior Geospatial Data Scientist, Gijs van den Dool to discuss Common Space’s work focusing on building open-licensed, freely accessible, high-resolution earth observation satellites dedicated to humanitarian aid. They discuss the project from technical vantage points and address core problems that Common Space aims to solve. They touch on the accessibility of satellites and data for use by humanitarian organizations and how aid funding cuts, structural changes in the commercial imagery market, limited observational capacity and competition combined with the critical need for the humanitarian community to avoid overreliance on third-parties for critical services, necessitate the development of initiatives like Common Space. Brent Phillips who produces the Humanitarian AI Today podcast incorporates a new question into the mini-series, asking Bill: If you were standing in front of a bold transformative philanthropist like MacKenzie Scott, what would be your argument for funding Common Space? Bill’s answer outlines the importance of providing the humanitarian community with guaranteed access to satellite imagery. Substack notes: https://humanitarianaitoday.substack.com/p/bill-greer-from-common-space-on-building

Voices is a new mini-series from Humanitarian AI Today. In daily five-minute flashpods we pass the mic to innovators, researchers and practitioners on the humanitarian front lines, delivering real-time news on how they are building, testing and collaborating on uses of artificial intelligence. In this flashpod, Rich Woods, Global Fundraising Lead with Tech To The Rescue, joins Brent Phillips, Humanitarian AI Today Producer, to discuss the impact of artificial intelligence on fundraising and how Tech To The Rescue is adapting and leveraging AI while prioritizing authentic human interaction with grantmakers. They discuss Tech To The Rescue’s mission and evolution and speak in depth about how AI is challenging fundraisers. Rich emphasizes that while AI allows fundraisers to conduct deep research and personalize outreach on a massive scale, there is a significant risk of losing authenticity. Prospecting can generate vast amounts of information quickly, but the fundraiser may lack a genuine connection to the data. He stresses the importance of taking the time to "live" the research to ensure communications remain human-to-human. Looking toward systemic changes, Rich shares his hope that AI can help reform the fundraising process, which he calls a long-broken and resource-heavy system for nonprofits. Peering further into the future, Rich envisions AI applications acting as matchmakers, connecting funders and organizations with shared interests to facilitate open, valuable conversations and partnerships. Substack notes: https://humanitarianaitoday.substack.com/p/rich-woods-from-tech-to-the-rescue

Voices is a new mini-series from Humanitarian AI Today. In daily five-minute flashpods we pass the mic to innovators, researchers and practitioners on the humanitarian front lines, delivering real-time news on how they are building, testing and collaborating on uses of artificial intelligence. In this flashpod, Shannon Farley, Co-founder and Executive Director of Fast Forward joins Eric Talbert, Co-founder of MedCycle Network to discuss Fast Forward’s 2025 AI for Humanity Report, which is a roadmap for harnessing AI for impact written by Fast Forward with support from Google. For over a decade, Fast Forward has supported social good initiatives, evolving with technology to become an important and direct accelerator of AI adoption in the social sector. Drawing upon its experience, the 2025 AI for Humanity report shows how nonprofits are using AI to transform lives and offers a roadmap for how to build on these successes responsibly. However, while AI has been able to create incredible efficiencies and opportunities, it also comes with new costs. The report reveals that a lack of funding is the most common obstacle for nonprofits, often preventing them from hiring the specialized tech experts they need to move uses of AI forward. Shannon emphasizes that philanthropy can directly address this gap. Even modest budget increases allow nonprofits to dramatically expand their reach, meaning an investment in a nonprofit's AI talent and infrastructure is a direct investment in scaling social impact. Substack Notes: https://open.substack.com/pub/humanitarianaitoday/p/shannon-farley-on-fast-forwards-2025?r=e9cbk&utm_campaign=post&utm_medium=web&showWelcomeOnShare=false

Voices is a new mini-series from Humanitarian AI Today. In daily five-minute flashpods we pass the mic to innovators, researchers and practitioners on the humanitarian front lines, delivering real-time news on how they are building, testing and collaborating on uses of artificial intelligence. In this flashpod, Eric Talbert, Co-founder of MedCycle Network joins Humanitarian AI Today host Brent Phillips to discuss Eric’s background in human rights, supply and logistics, MedCycle’s work, their partnership with the Hellman Foundation and ways MedCycle is experimenting with artificial intelligence. MedCycle Network facilitates the collection of donated, high-quality surplus medical supplies and equipment so they can be distributed to local safety net clinics. Eric and Brent discuss challenges that organizations like MedCycle and similar organizations like MedShare and Partners for World Health share around working with hospitals and companies to prevent medical supplies from being discarded that could otherwise be donated to needy healthcare providers struggling to care for their communities. They talk how AI-powered decentralized surplus supplies ecosystems can impact and inform choices made around the trash barrel in real-time, to redirect usable supplies to clinics in need on an item-by-item basis. Substack Notes: https://open.substack.com/pub/humanitarianaitoday/p/eric-talbert-from-medcycle-networks?r=e9cbk&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true

Voices is a new mini-series from Humanitarian AI Today. In daily five-minute flashpods we pass the mic to innovators, researchers and practitioners on the humanitarian front lines, delivering real-time news on how they are building, testing and collaborating on uses of artificial intelligence. In this episode, Philippe Stoll, Senior "Techplomacy" Delegate with the International Committee of the Red Cross (ICRC), joins Humanitarian AI Today host Brent Phillips to discuss the complex issues that digital technologies create for humanitarian organizations and for people affected by conflict from an IT perspective. They discuss the growing pressure on organizations to experiment with AI, highlighting the significant backend IT effort required to safely deploy and maintain these systems. This deployment introduces new operational and security risks, demanding a highly cautious and ethical "do-no-harm" approach to protect vulnerable populations. Stoll also explains how the ICRC collaborates with academia to help evaluate new applications and find solutions to complex problems. Philippe closes with a call for greater cross-disciplinary collaboration, urging experts from humanitarian, academic, and technology sectors to engage with one another to better understand each other's perspectives. Notes: https://humanitarianaitoday.substack.com/p/philippe-stoll-on-ai-techplomacy

Voices is a new mini-series from Humanitarian AI Today. In daily five-minute flashpods we pass the mic to innovators, researchers and practitioners on the humanitarian front lines, delivering real-time news on how they are building, testing and collaborating on uses of artificial intelligence. On this Voices flashpod, Perry Hewitt, Chief Strategy Officer at Data.org joins Humanitarian AI Today podcast host Brent Phillips to discuss Data.org’s role in supporting the social impact community’s transition into digital and its transition now into data and AI, and Data.org’s new report ‘Accelerate What’s Possible’ - a roadmap for how data and AI can be used innovatively and intentionally to drive social impact. They also touch on Data.org’s Activate AI: Economic Opportunity Challenge, their partnership with Zoom and the next Accelerate conference taking place in June 2026 in Bogotá, Colombia on the Javeriana University campus. Notes: https://humanitarianaitoday.substack.com/p/perry-hewitt-from-dataorg-on-accelerating

Voices is a new mini-series from Humanitarian AI Today. In daily five-minute flashpods we pass the mic to innovators, researchers and practitioners on the humanitarian front lines, delivering real-time news on how they are building, testing and collaborating on uses of artificial intelligence. In this Voices flashpod, Pam Boiros, a Founding Member of Women Applying AI speaks with Sandra Uwantege Hart, a Strategist focusing on Inclusive Innovation and Humanitarian Action from PoliSync, and Brent Phillips, Humanitarian AI Today podcast producer, about the launch of Women Applying AI, a free membership-driven community bringing more women into the field of Artificial Intelligence at all levels of experience. They touch on how AI can empower women, particularly non-technical female founders who can use AI to tackle business, life or world problems. They discuss the AI gender gap, the need for supportive initiatives and mentors for women, and the ways AI can empower female innovators. They also discuss Boston AI Week and touch on MIT’s project NANDA. Notes: https://open.substack.com/pub/humanitarianaitoday/p/pam-boiros-on-women-advancing-ai?r=e9cbk&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true

Voices is a new mini-series from Humanitarian AI Today. In daily five-minute flashpods we pass the mic to innovators, researchers and practitioners on the humanitarian front lines, delivering real-time news on how they are building, testing and collaborating on uses of artificial intelligence. In this Voices flashpod, Technology Strategist Sandra Uwantege Hart from PoliSync joins Humanitarian AI Today podcast host Brent Phillips to explore inclusive uses of emerging technologies like AI in humanitarian action. Their conversation covers PoliSync’s work bringing technical experts together to explore emerging issues in humanitarian aid, the use of AI to empower local organizations and startups, and the broader convergence of technologies we’re seeing today. Recorded against the backdrop of the UN General Assembly, they also examine the state of the humanitarian sector and the pressure on large international agencies to innovate for greater efficiency. Sandra closes the discussion with a call for improving digital access and literacy to help people help themselves, warning that old practices could perpetuate inequality if the sector doesn't engage correctly with new technology. Notes: https://humanitarianaitoday.substack.com/p/sandra-hart-from-polisync-on-engaging

Voices is a new series of bite-sized interviews produced by Humanitarian AI Today. Every day we pass the mic to a new voice from the humanitarian AI community to hear about new projects, research and advancements shaping our field. In this pilot episode of Voices, Suzy Madigan speaks with the podcast's producer about her new work supporting the CDAC Network (Communicating with Disaster-Affected Communities). Against the backdrop of the UN General Assembly, they also discuss how aid cuts and competition for funds are driving aid agencies to seek out AI ‘efficiencies’, creating new risks for humanitarian organizations and communities unless effective safeguards are put in place. The Machine Race: https://themachinerace.substack.com/ Episode notes: https://humanitarianaitoday.substack.com/p/suzy-madigan-on-ai-aid-reform

This 100th Humanitarian AI Today episode focuses on blind spots in AI safety and aligning AI with International Humanitarian Law. Guest host, Andre Heller, Director of Signpost at the International Rescue Committee (IRC), speaks with Sergio Coronado, Chief Information Officer with NATO’s Support and Procurement Agency (NSPA), about important research that he is heading at the Luxembourg Tech School studying “blind spots” in AI safety at the intersection of artificial intelligence and International Humanitarian Law (IHL). Dr. Coronado speaks in detail about his team's groundbreaking research, which tested leading AI models against codified rules of humanitarian law. The conversation delves into the chilling discovery that while models refuse obviously harmful requests about 90% of the time, they can still for example be prompted to generate malicious code for targeting civilian infrastructure like hospitals, contrary to IHL. This dialogue moves beyond identifying the problem to explore tangible solutions, highlighting how simple interventions can dramatically improve AI's adherence to legal principles. It serves as a powerful call to action for the humanitarian and technology communities to bridge this dangerous gap and champion the development of AI that is not just powerful, but principled and fundamentally law-adherent. Interview Notes: https://medium.com/humanitarian-ai-today/sergio-coronado-on-blind-spots-in-ai-safety-and-international-humanitarian-law-40b64590a119

In a world where AI-powered question answering interfaces and AI agents are becoming the new informational gatekeepers, how can humanitarian organizations adapt their communications strategies to stay visible, credible, and prominent? In this episode of Humanitarian AI Today, guest host Roderick Besseling, Head of the Data and Analytics Unit at the Norwegian Refugee Council, speaks with Matthew Brown from Profound, a startup that helps companies track, control, and optimize their marketing and communications content for the agentic internet. Joined by Lucy Hall, a Data and Evidence Specialist from Save the Children's Humanitarian Leadership Academy and Brent Phillips, Roderick and Matt discuss a critical challenge facing the humanitarian sector caused by artificial intelligence upending the world of search, simultaneously disrupting the industry and transforming the very nature of how we access information. This disruption forces a pivotal choice: organizations must adapt their communication strategies, or risk becoming invisible to the donors and communities they serve. Matt explains how Profound helps companies and organizations analyze their "AI visibility" by tracking how, when, and in what context their brand is mentioned by question-answering interfaces like ChatGPT, Gemini and Perplexity. He explains how Answer or Agentic Engine Optimization (AEO) works and how Profound can help organizations learn to generate the high-quality, semantically rich, and well-structured content that AI agents favor, ensuring that their communications are not just seen, but are recognized as trustworthy and reliable. The conversation also explores how AEO can support the localization agenda within the humanitarian sector. Matt argues that this technological shift can "level-up the playing field," giving local and grassroots organizations a better chance at visibility. Because AEO prioritizes well-structured, helpful content over large budgets and traditional SEO tactics, smaller organizations with less resources have a new opportunity to be discovered, ensuring their vital work is visible to donors and partners from the community level all the way up to large UN agencies. Episode notes: https://medium.com/humanitarian-ai-today/matthew-brown-from-profound-on-agentic-search-engine-optimization-aeo-for-humanitarian-75ba0e6560c6

In this episode of Humanitarian AI Today, guest host Doug Smith, Acting CEO of Data Friendly Space, speaks with Pradyumna Chari, a postdoctoral associate at the MIT Media Lab about Project NANDA. This initiative is building the foundational layer for an internet of AI agents through a broad coalition of academic institutions, major technology corporations, specialized AI startups, and the open-source community. Pradyumna explains how components like the NANDA Index create a "handshake layer" for intelligent agents to discover, coordinate, and transact with each other. This system is designed to shape the future of knowledge sharing, enabling agents to transact in privacy-preserving "intelligence" and "insights" rather than raw data. Doug and Pradyumna explore how this unlocks the potential for a "mesh" of interconnected agents to revolutionize humanitarian response. With this technology in a formative stage—much like the early World Wide Web—the humanitarian community has a critical opportunity to help shape its infrastructure. Tune in to learn how your organization can get involved and ensure this powerful new ecosystem is built to meet humanitarian needs from the ground up. Episode notes: https://humanitarianaitoday.medium.com/pradyumna-chari-introduces-project-nanda-to-humanitarian-organizations-333478f5f049

Humanitarian AI Today, guest host Mala Kumar, Head of Impact at Humane Intelligence, sits down with Andre Heller, Director of Signpost at the International Rescue Committee (IRC). They discuss Signpost's recent research paper on piloting an "information assistant," detailing the technical architecture, evaluation methods, and lessons learned from the project The conversation also addresses the significant challenges facing the sector, including a funding crisis that has impacted the pace, scale, and scope of critical research being carried out across the humanitarian community advancing humanitarian applications of artificial intelligence. In the midst of this crisis and the explosive growth of AI, Andre and Mala emphasize the need for more rigorous, scientific, and collaborative approaches to AI development and evaluation. They speak in detail about open source AI and what it means to the humanitarian community. And Mala explains how her organization, Humane Intelligence, is working to professionalize AI evaluation by creating a community of practice around fair and representative algorithmic auditing. She describes their work in red teaming, conducting "bias bounties," and the future development of open-source "evaluation cards" to make evaluation methodologies more transparent and reusable. Episode notes: https://medium.com/humanitarian-ai-today/andre-heller-and-mala-kumar-discuss-signposts-pilot-ai-assistant-903b5b8ab787