From false attribution to synthetic personation: personational integrity and a residual tort for deepfake harms

Abstract

Deepfakes do not stand categorically outside defamation law. When synthetic media communicate a false factual attribution about an identifiable person and are understood in that way, ordinary defamation doctrine may provide a remedy. The harder cases are residual ones in which serious injury or interference persists even though the synthetic nature of the representation is known, no reasonable audience treats the represented event as authentic, or the central wrong is identity substitution rather than reputational belief. This paper reframes the legal problem around that distinction and identifies personational integrity as the distinct interest at stake: the interest of a living person in not having an identifiable embodiment of that person materially deployed to occupy the person’s role as speaker, actor, participant, endorser, or decision-maker in a seriously wrongful fabricated context. Personational integrity is narrower than a general right to control one’s image and is analytically distinct from reputation, privacy, publicity value, and emotional tranquility, although violations may also injure those interests. Building on this account, the paper proposes a technology-neutral tort of Synthetic Personation Harm (SPH). Liability requires materially fabricated personation, identification, lack of authorization, contextual wrongfulness, serious interference with personational integrity or substantial consequential harm, causation, and an appropriate fault showing; non-consent alone is insufficient. Expressive safeguards for news, education, historical representation, parody, satire, and transformative art are built into the framework, together with an anti-circumvention rule preventing SPH from being used merely to repackage a defamation claim. A revised decision algorithm does not depend on a deepfake-detector threshold, and an ordinal harm matrix separates the core personational interference from consequential losses. Applications to the Zelenskyy surrender deepfake, the 2024 Hong Kong video-conference fraud, a labeled intimate fabrication, and educational and parodic boundary cases illustrate overlap, residuality, and protected use.