Lu on Data Privacy, Human Rights, and Algorithmic Opacity

Sylvia Lu (UC Berkeley School of Law) has posted “Data Privacy, Human Rights, and Algorithmic Opacity” (California Law Review, Vol. 110, 2022) on SSRN. Here is the abstract:

Decades ago, it was difficult to imagine a reality in which artificial intelligence (AI) could penetrate every corner of our lives to monitor our innermost selves for commercial interests. Within a few decades, the private sector has seen a wild proliferation of AI systems, many of which are more powerful and penetrating than anticipated. In many cases, machine-learning-based AI systems have become “the power behind the throne,” tracking user activities and making fateful decisions through predictive analysis of personal information. However, machine-learning algorithms can be technically complex and legally claimed as trade secrets, creating an opacity that hinders oversight of AI systems. Accordingly, many AI-based services and products have been found to be invasive, manipulative, and biased, eroding privacy rules and human rights in modern society.

The emergence of advanced AI systems thus generates a deeper tension between algorithmic secrecy and data privacy. Yet, in today’s policy debate, algorithmic transparency in a privacy context is an issue that is equally important but managerially disregarded, commercially evasive, and legally unactualized. This Note illustrates how regulators should rethink strategies regarding transparency for privacy protection through the interplay of human rights, disclosure regulations, and whistleblowing systems. It discusses how machine-learning algorithms threaten privacy protection through algorithmic opacity, assesses the effectiveness of the EU’s response to privacy issues raised by opaque AI systems, demonstrates the GDPR’s inadequacy in addressing privacy issues caused by algorithmic opacity, and proposes new algorithmic transparency strategies toward privacy protection, along with a broad array of policy implications and suggested moves. The analytical results indicate that in a world where algorithmic opacity has become a strategic tool for firms to escape accountability, regulators in the EU, the US, and elsewhere should adopt a human-rights-based approach to impose a social transparency duty on firms deploying high-risk AI techniques.