Charvi Rastogi, a Senior Research Scientist at Google DeepMind, will give a talk on Friday, November 6, at 1:00 PM in Tyler 106 titled “Going PLACES: Participatory Localized Red Teaming for Text-to-Image Safety in the Global South.” According to the abstract, text-to-image (T2I) models are globally deployed, but their safety frameworks are largely calibrated to a Western-centric default, creating vulnerabilities for the rest of the world. To embrace cultural pluralism and bring historically under-represented perspectives into T2I safety, the work conducts localised community-centered red teaming studies in the Global South, prioritizing localization and participation by focusing on secondary urban centers and running community engagement and training workshops to contextualize local norms. The result is PLACES, a dataset comprising over 26,000 examples of T2I model failures collected in partnership with universities in Ghana, Nigeria, and two regions of India (Karnataka and Punjab). Analysis reveals diversity in socio-cultural and linguistic attributes compared with existing geography-agnostic crowdsourced red-teaming data, unique adversarial patterns enabled by local cultural and linguistic nuances, and distinct clusters within region around specific themes, such as religion in India. The work also uncovers structural contextual gaps in existing safety frameworks by identifying novel harms showing normative dissonance, such as violating religious norms, ignoring local customs, and ominous symbolism. It argues that expanding T2I safety requires moving beyond mere scale to