Reconciling Deepfake Regulations vis-à-vis Human Rights: A Comparative Legal Analysis Across Jurisdictions

Abstract

Synthetic media produced by generative artificial intelligence, commonly called deepfakes, have become an industrialised instrument of financial fraud, electoral manipulation, and image based sexual abuse, and they now threaten the evidentiary foundations of courts and human rights investigations. This Article assesses how four jurisdictions, the United States, the United Kingdom, the European Union, and India, had addressed the problem by mid 2026, and it measures each regime against the International Covenant on Civil and Political Rights (ICCPR), in particular the guarantees of a fair trial in Article 14, privacy in Article 17, expression in Article 19, and political participation in Article 25. The comparison reveals four distinct regulatory philosophies: rights constrained, state-led regulation in the United States; upstream criminalisation of the technology supply chain in the United Kingdom; transparency-based, risk-tiered governance in the European Union; and accelerated intermediary liability in India. Each protects some rights at the expense of others, and all depend on provenance technology whose metadata is easily stripped and whose cryptographic assurances are weaker than they appear. The Article also examines the “authentication deficit” confronting the International Criminal Court. It concludes that domestic statutes alone cannot adequately address a borderless technical problem and proposes an additional protocol under the United Nations Global Digital Compact establishing “Provenance by Design,” combining mandatory preservation of provenance metadata by distribution platforms with the formal adoption of the Berkeley Protocol by international tribunals, and disciplined throughout by the necessity and proportionality requirements of Article 19(3) of the ICCPR.