An Afrocentric Perspective on Algorithm Watermarking of AI-generated Content.

ICLR 2025 Workshop BuildingTrust Submission145 Authors

Track: Long Paper Track (up to 9 pages)

Keywords: Watermark, trust, fairness

TL;DR: An afro-centric perspective on algorithm watermarks beyond technical metrics

Abstract: Digital-driven misinformation, counterfeiting, and copyright violations have become a growing concern in Africa. The prevalence of Artificial intelligence content (AIGC) has the potential to widen its impact and create more challenges for the people on the continent. AIGC poses a dual challenge. First, creatives who have worked so hard to create a masterpiece see their work being illegally duplicated or used without their consent. The other unsuspecting individuals have fallen prey to misinformation caused by AIGC. The reason, amongst many, could be the regulatory gaps in the law governing data protection, copyright and even artificial intelligence. This paper argues that curating technical watermarking methodologies/techniques is insufficient, considering the uniqueness of the African continent. It further addresses the regulatory gaps by examining the existing laws and proposing an Afrocentric perspective on AIGC using Nigeria, Kenya, Egypt and South Africa as case studies.

Submission Number: 145