This paper explores how algorithmic systems reconfigure established notions of the image, reframing it as fragmented, actionable data units rather than a coherent representational form, thereby raising critical questions about privacy rights such as the Right to Be Forgotten. Drawing on the example of the Swedish police’s use of facial recognition technology, it illustrates how photographs are reconstituted as actionable elements within predictive and classificatory infrastructures. As archives become computational and machine-driven, established understandings of deletion, autonomy, and accountability are increasingly destabilised. The analysis shows that the conceptual and regulatory limits of erasure become particularly salient in contexts where data is no longer merely stored but persistently circulated and acted upon. The paper concludes that legal frameworks premised on coherent data subjects and discrete visual artefacts are inadequate to the distributed logics of algorithmic governance, and it calls for a reimagining of archival and data governance practices that recognise the enduring operational life of images beyond their initial capture.