The Responsible AI Lab: What can we learn from 5 real AI failures in the impact sector? (practical workshop)

Responsible AI principles are relatively easy to write. The real test comes when AI meets an organization: imperfect data, stretched teams, unclear accountability, pressure to move quickly, and consequences nobody anticipated until something goes wrong. What do the biggest AI failures of recent years have in common? Someone, somewhere, didn’t see it coming. The Responsible AI Lab is an immersive learning experience for social and climate changemakers who want to get beyond principles and examine what responsible AI requires in practice. Instead of starting with hypothetical risks, the Lab will start with failures that have already happened. Participants will unpack real cases where AI systems produced harmful, biased, unsafe, or simply unintended outcomes and work backwards. What went wrong? Who could have spotted it? At what point could someone have intervened? Was the problem the technology itself, or the team, governance, incentives, assumptions, and processes surrounding it? The harder question is what those failures reveal about our own organizations. As nonprofits, philanthropies, social enterprises, climate organizations, and development institutions begin experimenting with and deploying AI, responsibility cannot sit solely with a technical team or an AI policy document. Organizations need ways to decide where AI should be used, identify risks before deployment, establish meaningful accountability, involve the right people in decisions, and recognize when a seemingly efficient tool could create consequences for the communities they serve. This Lab creates space to test those questions before the crisis happens. What to expect Part case study, part hands-on workshop, and part candid peer conversation. Over approximately two hours, participants will work through real AI failures, identify where safeguards broke down, and translate those lessons into their own organizational context. There will be no slide marathon. Expect small-group problem-solving, practical exercises, peer learning, and honest discussion about where responsible AI gets difficult once principles collide with budgets, deadlines, organizational politics, and real-world implementation. Participants will also have time to compare approaches with peers working across social impact, humanitarian response, development, and climate, making the Lab as much about learning from one another as learning from the cases themselves. Who this is for This session is for social and climate changemakers who are already using AI or expect to be responsible for how their organizations use it, including nonprofit and philanthropic leaders, humanitarian and development practitioners, program and operations teams, technology leaders, and others responsible for turning responsible AI principles into actual organizational practice. What you will get out of it Hosted by