When AI Remembers Too Much: Reinventing the Right to Be Forgotten for the Generative Age

Abstract

The emergence of generative artificial intelligence (AI) systems poses novel challenges for the right to be forgotten. While this right gained prominence following the 2014 Google Spain v. Gonzalez case, generative AI’s limitless memory and ability to reproduce identifiable data from fragments threaten traditional conceptions of forgetting. This article traces the evolution of the right to be forgotten from its privacy law origins towards an independent entitlement grounded in self-determination for personal information. However, it contends the inherent limitations of using current anonymization, deletion, and geographical blocking mechanisms to prevent AI models from retaining personal data render forgetting infeasible. Moreover, the technical costs of forgetting - including tracking derivations and retraining models - could undermine enforceability. Therefore, this article advocates for a balanced legal approach that acknowledges the value of the right to forget while considering the constraints of implementing the right forgenerative AI. Although existing frameworks like the European Union’s GDPR provide a foundation, continuous regulatory evolution through oversight bodies and industry collaboration is imperative. This article underscores how the right to be forgotten must be reconceptualized to address the reality of generative AI systems. It provides an interdisciplinary analysis of this right’s limitations and proposes strategies to reconcile human dignity and autonomy with the emerging technological realities of AI. The article’s original contribution lies in its nuanced approach to integrating legal and technical dimensions to develop adaptive frameworks for the right to be forgotten in the age of generative AI.

Keywords: Generative AI, Right to be forgotten, GDPR, Personal data, Privacy, Data protection, AI memory, Self-determination, Ethical AI, AI Regulation, AI Ethics, ChatGPT