Deceptive Deepfakes: Is the Law Coping with AI-Altered Representations of Ourselves?

Deceptive deepfakes are AI-generated hyper-realistic content that harms individuals (e.g. pornographic deepfake content) or society (as disinformation and impersonation tools). The technologies are becoming increasingly sophisticated, making their detection extremely difficult. Deepfakes can impair national and international security, undermine elections, threaten democracy, and challenge justice, but also cause mental distress and reputational damage. What are the regulatory answers to these deceptive deepfakes? As this paper will show, there is no single approach and answer to the topic. Currently, the problem is tackled by existing, albeit non-specific regulations, (e.g. privacy, intellectual property, image rights), and by newly adopted regulations focusing on a specific aspect (deepfakes aimed at manipulating elections or non-consensual pornographic materials). To present an overview of these regulations, the paper compares the approaches of the USA and the EU, including examples of national legislation. Summarizing the findings, the paper shows the laws need to regulate the ecosystem of actors involved in these deceptive contents (from the creators to the distributors and possibly the viewers in case of deepfake child pornography). Beyond the ecosystem, solutions should also focus on individuals who are victims of these deepfakes created without their consent and knowledge. A different approach might be well needed to ensure they can defend their digital self and prevent malicious AI-altered reproductions and use of themselves.