Recreating historical figures on screen: the ethics of reshaping history through AI resurrections

When bringing past events to the screen for a broad audience, archival footage often falls short of providing a complete picture. To bridge these gaps, historical documentaries and educational programs have traditionally relied on dramatic recreations and explanatory animations. However, the «digital resurrection» of deceased individuals using audio and video deepfakes is rapidly transforming the field. By training AI on archival material, filmmakers can now realistically reconstruct historical figures, opening new possibilities for historical storytelling and engagement. This technological advancement, however, raises complex ethical questions regarding authenticity, representation, and the very nature of historical memory. This presentation explores both the transformative potential and the ethical complexities of this evolving technology, focusing on its impact on our understanding and interaction with the past.

Specifically, this study examines the implications of deepfake technology —the synthesis of realistic yet artificial human likenesses— for key stakeholders, including archivists, filmmakers, and the public. Case studies, including the AI-cloned voice in Roadrunner: A Film About Anthony Bourdain (2021) and the digitally resurrected celebrities in Hôtel du Temps (2022–2024), highlight diverse approaches and the ethical debates they provoked. These are contrasted with the ethically-grounded recreation of comedian Pepe Rubianes in The World of Pepe Rubianes (2024), which prioritized family consent and audience transparency, illustrating a spectrum of practice and the crucial role of ethical considerations.

This research further investigates the broader impact of generative AI on historical narratives. While acknowledging its potential to enrich understanding by creating a seemingly tangible connection to the past, it also emphasizes the risks of misrepresentation and manipulation. The presentation argues that responsible implementation is paramount and must be guided by clear ethical frameworks. It analyzes emerging principles from organizations like the Archival Producers Alliance and media outlets such as the BBC, alongside academic proposals from the UB Chair in Artificial Intelligence and Media. These guidelines converge on the need for transparency, accuracy, the prioritization of primary sources, and the respectful representation of deceased individuals to protect both historical integrity and personal rights.