An Art-centric perspective on AI-based content moderation of nudity

TL;DR

This paper examines algorithmic censorship of artistic nudity online, revealing biases and limitations in current classifiers, and proposes a multi-modal zero-shot approach to improve detection accuracy.

Contribution

It uncovers gender and stylistic biases in existing classifiers and introduces a novel multi-modal zero-shot method for better artistic nudity detection.

Findings

Existing classifiers show gender bias.

Visual-only models have technical limitations.

Multi-modal approach improves classification accuracy.

Abstract

At a time when the influence of generative Artificial Intelligence on visual arts is a highly debated topic, we raise the attention towards a more subtle phenomenon: the algorithmic censorship of artistic nudity online. We analyze the performance of three "Not-Safe-For-Work'' image classifiers on artistic nudity, and empirically uncover the existence of a gender and a stylistic bias, as well as evident technical limitations, especially when only considering visual information. Hence, we propose a multi-modal zero-shot classification approach that improves artistic nudity classification. From our research, we draw several implications that we hope will inform future research on this topic.