Every piece of data that is made more accessible to the public is now at risk of consumption by large language models. Writers, artists, and other cultural custodians are now grappling with the consequences of having their data stolen and repurposed without consent, credit or compensation. As we crest the Peak of Inflated Expectations of this AI hype cycle, we must understand that databases and other archives are not just technical systems, but political and cultural infrastructures that enable capture.

Long-term, we need better legislation that protects humans and our right to cultural expression, privacy, and data sovereignty. In the meantime, we must reimagine protocols and access to data sources that centre relational responsibility. This talk explores what needs to be true for database architectures to carry culture without flattening it—without imposing imperial violence and extractive logics upon its contents—and the role of hyperlocal, community-based data stewardship.