Pelin Kasar

Autonomous technologies, particularly self-learning AI systems, are often said to create responsibility gaps—cases where harm is caused, yet no one is responsible, because no one appears to meet the control and epistemic conditions typically required for moral responsibility. In this paper, I argue that this problem is better understood as a challenge of attributing moral responsibility for unintentional actions. I suggest that the unintended, harmful AI-based outcomes should be characterized as unintentional actions that can be traced back to human agents. On this basis, I argue that while such actions may be unintentional under some description—and thus potentially excusable—they do not negate moral responsibility. Instead, they modify it: designers and users remain responsible due to the moral residue left by their involvement, and they may bear reparative obligations, such as offering explanations, apologies, or compensation for the harm caused. In high-stakes cases, moral agents may still need to take responsibility—and may, in some contexts, be appropriate targets of blame.