AI and the Inner Sphere: Privacy and Proportionality under International Human Rights Law

Sammanfattning

This thesis examines how contemporary artificial intelligence (AI) systems challenge the protection of the right to privacy under international human rights law, particularly Article 17 of the International Covenant on Civil and Political Rights and Article 8 of the European Convention on Human Rights. It argues that privacy should not be understood only as protection against unlawful access to personal information, but as a legal safeguard for the individual's inner sphere, including autonomy, mental integrity and identity formation. The central argument in this thesis is that AI systems interfere with privacy in ways that go beyond traditional forms of data collection or disclosure. Through behavioral inference, predictive analytics, emotional profiling, digital phenotyping, personalized recommendation systems, and conversational AI, such systems increasingly shape the environments within which individuals think, decide, behave and develop identity. These interferences are often not visible, immediate, or easily measurable. Instead, they operate through psychological pressure, chilling effects, emotional dependency, and gradual behavioral modification. Therefore, existing proportionality analysis under international privacy law remains under-inclusive when it fails to recognize cumulative, psychological, and structurally embedded harms as legally relevant interferences with privacy. Although the existing doctrine already contains important mechanisms, including legality, foreseeability, necessity, safeguards, fair balance, and positive obligations, its current application remains too closely attached to model of salient harm. This model is better suited for visible, direct, and procedurally identifiable privacy violations than to the slow and diffuse harms produced by AI-mediated environments.