The ethics of deepfake victim impact statements at sentencing

Abstract In 2025 a deepfake of Chris Pelkey, a murder victim, was submitted as a victim impact statement (VIS) at a court in Arizona. This paper considers the ethical issues surrounding the new phenomenon of AI-generated VIS at sentencing. It argues that objections to deepfakes that are based on the intention principle or which are motivated by the fear that deepfake VIS may unduly influence sentencing should be rejected or, at least, has a considerably narrower scope than friends of such objections may believe. Moreover, it argues that while some types of deepfake VIS will struggle to serve some the functions that VIS are intended to serve, this is far from the case for all deepfake VIS. Indeed, the paper finds, surprisingly, that in some cases deepfake VIS are potentially a better way to meet the intended goals of VIS. This, it is argued, provides support for the view that in some cases it is morally desirable for deepfake VIS to be used at sentencing.