If the digital cultural heritage community is to play a meaningful role in the development of AI, it must first develop a shared vision. Yet the vague use of terms such as ethical, trustworthy and participatory AI, combined with the lack of concrete practices to make these concepts tangible, creates complexity and uncertainty. As the Europeana Initiative and its partners scale up the common European data space for cultural heritage, the question becomes urgent: what role should the data space take in responsible AI development and adoption? This paper is the first outcome of the “Alignment Assembly on Culture for AI”, a collective intelligence exercise initiated within the data space, which engaged circa 400 professionals. It maps the results of the community consultation, identifies areas of consensus, friction and uncertainty and pinpoints topics that require further exploration. The paper identifies two contrasting opinion groups: boundary-setters (32% of respondents), critical of AI use, and opportunity-seekers (68%), advocating for its adoption. It explores key themes and insights that emerged from this exercise - ranging from defining and defending the sector’s position and responsibilities, to positioning heritage data as responsible AI training material, using AI in a way that is climate responsible or preventing the perpetuation of biases when AI is used with heritage data, among others. Building on these insights, the paper outlines directions for further community work and proposes a set of priorities, an action plan, and immediate next steps. The Alignment Assembly thus serves as the starting point for a deeper, ongoing dialogue within the digital cultural heritage community about its engagement with and stance on AI technologies. We also provide the data from the survey publicly launched on Pol.is for the purpose of this paper.