What happens when machines are trained to care? From wellness platforms that track mood and behaviour, to therapeutic chatbots, classroom emotion-recognition software, and the AI girlfriend/boyfriend experience, today’s algorithmic tools increasingly promise intimacy, attentiveness, and support—while simultaneously reorganizing perception, flattening sensation, and quietly hardwiring normative models of emotion, embodiment, and subjectivity. As these systems rush to patch the widening holes in precarious care infrastructures, they do so by running on hidden circuits of energy and data, tied to ever-intensifying regimes of planetary extraction and the spectre of human labour that props up so-called artificially intelligent systems. To figure care in the age of AI is thus to ask how today’s “feeling machines” sense and standardize intimacy, how affective labour is black-boxed into protocols of learning, and how work, energy, and desire are recoded to naturalize some forms of care as authentic, others as artificial. Drawing on recent research-creation projects, the talk weaves theoretical inquiry, artistic experimentation, and pedagogical reflection to investigate care as a site of tension and ambivalence—where intimacy meets infrastructure, tenderness becomes protocol, and collective resistance might be reworked with and beyond the machine.