Explainable Artificial Intelligence (XAI) methods face significant challenges in evaluating their effectiveness, particularly when dealing with static and evolving data. A recent study highlights the limitations of XAI through the example of the DetoxAI image recognition system, which aims to detect bias and unlearn concepts1. The research underscores the need for more rigorous evaluation methods, including human-grounded assessments, to ensure that XAI techniques are reliable and trustworthy. The study also explores approaches for adapting explanation methods to dynamic data, a crucial aspect of real-world applications. The lack of robust evaluation frameworks for XAI poses significant concerns, as flawed explanations can lead to misguided decision-making. This matters to practitioners because developing effective XAI methods is critical for building trust in AI systems, and a flawed evaluation process can have far-reaching consequences.