We investigate the question of whether (and if so why) creating or distributing deepfake pornography of someone without their consent is inherently objectionable. We argue that nonconsensually distributing deepfake pornography of a living person on the internet is inherently pro tanto wrong in virtue of the fact that nonconsensually distributing intentionally non-veridical representations about someone violates their right that their social identity not be tampered with, a right which is grounded in their interest in being able to exercise autonomy over (...) their social relations with others. We go on to suggest that nonconsensual deepfakes are especially worrisome in connection with this right because they have a high degree of phenomenal immediacy, a property which corresponds inversely to the ease with which a representation can be doubted. We then suggest that nonconsensually creating and privately consuming deepfake pornography is worrisome but may not be inherently pro tanto wrong. Finally, we discuss the special issue of whether nonconsensually distributing deepfake pornography of a deceased person is inherently objectionable. We argue that the answer depends on how long it has been since the person died. (shrink)

Unsupervised learning algorithms are widely used for many important statistical tasks with numerous applications in science and industry. Yet despite their prevalence, they have attracted remarkably little philosophical scrutiny to date. This stands in stark contrast to supervised and reinforcement learning algorithms, which have been widely studied and critically evaluated, often with an emphasis on ethical concerns. In this article, I analyze three canonical unsupervised learning problems: clustering, abstraction, and generative modeling. I argue that these methods raise unique epistemological and (...) ontological questions, providing data-driven tools for discovering natural kinds and distinguishing essence from contingency. This analysis goes some way toward filling the lacuna in contemporary philosophical discourse on unsupervised learning, as well as bringing conceptual unity to a heterogeneous field more often described by what it isnot(i.e., supervised or reinforcement learning) than by what itis. I submit that unsupervised learning is not just a legitimate subject of philosophical inquiry but perhaps the most fundamental branch of all AI. However, an uncritical overreliance on unsupervised methods poses major epistemic and ethical risks. I conclude by advocating for a pragmatic, error-statistical approach that embraces the opportunities and mitigates the challenges posed by this powerful class of algorithms. (shrink)

[Author note: There is a video explainer of this paper on youtube at the new work in philosophy channel (search for surname+deepfakes).] -/- It is widely thought that deepfake videos are a significant and unprecedented threat to our epistemic practices. In some writing about deepfakes, manipulated videos appear as the harbingers of an unprecedented _epistemic apocalypse_. In this paper I want to take a critical look at some of the more catastrophic predictions about deepfake videos. I will argue for three (...) claims: (1) that once we recognise the role of social norms in the epistemology of recordings, deepfakes are much less concerning, (2) that the history of photographic manipulation reveals some important precedents, correcting claims about the novelty of deepfakes, and (3) that proposed solutions to deepfakes have been overly focused on technological interventions. My overall goal is not so much to argue that deepfakes are not a problem, but to argue that behind concerns around deepfakes lie a more general class of social problems about the organisation of our epistemic practices. (shrink)

In this paper, I draw a distinction between two types of deepfake, and unpack the deceptive strategies that are made possible by the second. The first category, which has been the focus of existing literature on the topic, consists of those deepfakes that act as a fabricated record of events, talk, and action, where any utterances included in the footage are not addressed to the audience of the deepfake. For instance, a fake video of two politicians conversing with one another. (...) The second category consists of those deepfakes that direct an illocutionary speech act - such as a request, injunction, invitation, or promise - to an addressee who is located outside of the recording. For instance, fake footage of a company director instructing their employee to make a payment, or of a military official urging the populace to flee for safety. Whereas the former category may deceive an audience by giving rise to false beliefs, the latter can more directly manipulate an agent's actions: the speech act's addressee may be moved to accept an invitation or a summons, follow a command, or heed a warning, and in doing so further a deceiver's unethical ends. (shrink)

Essentialists understand authenticity as an inherent quality of a person, object, artifact, or place, whereas constructionists consider authenticity as a social creation without any pre-given essence, factuality, or reality. In this paper, we move beyond the essentialist-constructionist dichotomy. Rather than focusing on the question whether authenticity can be found or needs to be constructed, we hook into the idea that authenticity is an interactive, culturally informed process of negotiation. In addition to essentialist and constructionist approaches, we discuss a third, less (...) well-known approach that cannot be reduced to any of the two forms. This approach celebrates the authenticity of inauthenticity by amplifying the made. We argue that the value of authenticity lies not in choosing for one of these approaches, but in the degree to which the process of negotiating authenticity enables a critical formation of selves and societies. Authenticity is often invoked as a method of social control or a mark of power relations: once something is defined as authentic, it is no longer questioned. Emerging technologies—especially data-driven technologies—have the capacity to conceal these power relations, propel a shift in power, and dominate authentication processes. This raises the question how processes of authentication can contribute to a critical formation of selves and societies, against the backdrop of emerging technologies. We argue in favor of an interactionist approach of authenticity and discuss the importance of creating space in authentication processes that are increasingly influenced by technology as an invisible actor. (shrink)

Deepfakes are hyper-realistic but fabricated videos created with the use of artificial intelligence. In the context of psychotherapy, the first studies on using deepfake technology are emerging, with potential applications including grief counselling and treatment for sexual violence-related trauma. This paper explores these applications from the perspective of medical ethics and health law. First, we question whether deepfake therapy can truly constitute good care. Important risks are dangerous situations or ‘triggers’ to the patient during data collection for the creation of (...) a deepfake, and when deepfake therapy is started, there are risks of overattachment and blurring of reality, which can complicate the grieving process or alter perceptions of perpetrators. Therapists must mitigate these risks, but more research is needed to evaluate deepfake therapy’s efficacy before it can be used at all. Second, we address the implications for the person depicted in the deepfake. We describe how privacy and portrait law apply and argue that the legitimate interests of those receiving therapy should outweigh the interests of the depicted, as long as the therapy is an effective and ‘last resort’ treatment option, overseen by a therapist and the deepfakes are handled carefully. We suggest specific preventative measures that can be taken to protect the depicted person’s privacy. Finally, we call for qualitative research with patients and therapists to explore dependencies and other unintended consequences. In conclusion, while deepfake therapy holds promise, the competing interests and ethicolegal complexities demand careful consideration and further investigation alongside the development and implementation of this technology. (shrink)

The fast development of synthetic media, commonly known as deepfakes, has cast new light on an old problem, namely—to what extent do people have a moral claim to their likeness, including personally distinguishing features such as their voice or face? That people have at least some such claim seems uncontroversial. In fact, several jurisdictions already combat deepfakes by appealing to a “right to identity.” Yet, an individual’s disapproval of appearing in a piece of synthetic media is sensible only insofar as (...) the replication is successful. There has to be some form of identity between the content and the natural person. The question, therefore, is how this identity can be established. How can we know whether the face or voice featured in a piece of synthetic content belongs to a person who makes claim to it? On a trivial level, this may seem an easy task—the person in the video is A insofar as he or she is recognised as being A. Providing more rigorous criteria, however, poses a serious challenge. In this paper, I draw on Turing’s imitation game, and Floridi’s method of levels of abstraction, to propose a heuristic to this end. I call it the identification game. Using this heuristic, I show that identity cannot be established independently of the purpose of the inquiry. More specifically, I argue that whether a person has a moral claim to content that allegedly uses their identity depends on the type of harm under consideration. (shrink)