Personal data flows across digital technologies integrated into people's lives and relationships. Increasingly, these technologies include Generative AI. (How) should personal data flow into and out of GenAI models? We investigate how people experience personal data collection in GenAI ecosystems and unpack the enablers and barriers to governing their data. We focus on personal data collection by Meta, specifically Instagram, in line with their recent policy update on processing user data to train GenAI models. We conducted semi-structured interviews with 20 Latin American Instagram users, based in Europe and Latin America. We discussed the acceptability of their data flowing in and out of GenAI models through different scenarios. Our results interrogate power dynamics in data collection, the (inter)personal nature of data, and the multiple unknowns concerning data and their algorithmic derivatives. We pose provocations around feelings of powerlessness, reframing (inter)personal data, and encountering unknown data and algorithms through design.
https://dl.acm.org/doi/10.1145/3706598.3713504
The ACM CHI Conference on Human Factors in Computing Systems (https://chi2025.acm.org/)