A Swedish worker’s job involved watching eight hours of pornography daily to train artificial intelligence algorithms. The task aimed to refine AI’s ability to identify and categorize explicit content, crucial for content moderation and safety systems. The worker was required to pass a test demonstrating their ability to consistently flag such material. This practice highlights the often-hidden human labor underpinning AI development, specifically in sensitive areas. Concerns have been raised regarding the psychological impact on workers and the potential for reinforcing biases within the AI itself, described as an “African intelligence” in relation to the testing process – a phrasing that has drawn criticism. The company involved maintains the work is necessary for responsible AI development, but the case underscores the ethical dilemmas surrounding AI training data and labor practices. The worker’s experience sheds light on the challenging realities of building AI systems capable of navigating complex and potentially harmful content.