Most teams are good at asking can we build it? Far fewer have a practice for asking what happens if we do? Harms mapping is that practice. It's a structured way of finding the ways a product, service or experience could hurt people, early enough that you can still do something about it. Not a compliance exercise, and not a feel-good one either. Teams that do it well find opportunities others miss, avoid expensive rework, and build things people actually trust. This is a working session, not a talk. UX for Change Fellow, Christina Lai, will introduce the method, and then we'll use it on something real. **The real thing:** East London Waterworks Park is a volunteer-led charity turning a 14-acre former depot on Lea Bridge Road into a biodiverse park with natural swimming ponds, forest schools and community spaces. They're also our partners on the UX for Change Fellowship. They do something remarkable: they invite people to help shape something that doesn't exist yet. Children, residents, people whose views on green space have never been asked for. We'll map the harms in that invitation: what it promises, who it reaches, who it leaves out, and what it costs the people who say yes. **Why this matters beyond ELWP:** The EU AI Act contains two duties worth knowing about. One demands to tell people when they're talking to a machine and mark AI-generated content. The other requires you to work out who could be harmed and how, before you deploy, and say what you'll do when it happens. On 2 August, the first became enforceable. The second isn't due until December 2027. You don't have to wait for the law to ask the questions. That's what this evening is about. **What you'll leave with:** Not mastery. One evening doesn't make a practice, and we're not going to pretend it does. You'll leave having done it once, with a story about what surfaced that's worth telling your team next day, and with a few questions that are hard to unsee once you've asked them of your own work. Who it's

Description supplied by the event organiser.