After reviewing some experiences of AI use in social services in Catalonia due to its key role in developing the first AI strategy in Spain, our work adopts an ethics from below perspective through Participatory Action Research (PAR), engaging social workers, policymakers, technologists, developers, and social scientists to assess AI-based social services in Catalonia. Two key meetings under the umbrella of the AI FORA project illuminated the discourse, highlighting the importance of continuous, inclusive evaluation to maintain community relevance and integrity. Methodologically, we employed focus groups for a comprehensive exploration of the advantages and disadvantages of AI in social services, and a World Cafe to gain expertise and perspective around four central themes: (1) data-driven social services for resource allocation, (2) predictive analytics and early intervention, (3) evaluation and continuous improvement, and (4) stakeholder collaboration. The discussion highlights the necessity of continuous ethical reassessment and updates within organizations to maintain integrity, especially in the context of AI in social services, emphasizing the importance of aligning with evolving societal norms and technological advancements. Additionally, it emphasizes the role of organizational culture and digital literacy in adopting AI, advocating for a balanced approach that integrates technological innovation with human empathy and judgement, while addressing the challenges of ensuring ethical AI deployment through proactive, inclusive, and culturally sensitive practices.