Deepfake technologies have progressed at an incredible pace. Thanks to the advancements in AI and machine learning, it is has become easy and cost-efficient to replace one person’s likeness, voice or image with another producing credible results. Such technologies can have positive and helpful applications in the education, creative and technological sectors to name a few. On the other hand, they also raise several concerns as seen in recent times.

Policy makers worldwide are aware of these risks and are taking various initiatives. The EU, for example, recently enacted the AI Act, which is the first legislation worldwide to have introduced harmonised rules for AI systems. This Act contains some specific provisions for deepfakes, notably a general obligation to disclose the artificial origin of the content. This talk based on research conducted in Japan and UK will outline a comparative perspective from law, policy, technology and practice. In particular, the paper will address the current gaps and provide insights from different sectors in highlighting ways in which the growing issue of deepfakes can be tackled. In doing so, the paper will make reference to the developments from Japan and the policy initiatives in tackling deepfakes in both jurisdictions. The talk will conclude with reflections on what the future holds for deepfakes, regulation, and AI technologies more broadly.