In recent years, Artificial Intelligence (AI)-driven decision systems have been widely applied in domains such as credit scoring, insurance risk assessment, and health monitoring, where predictive accuracy is critical. While these systems enhance decision-making capabilities, they also introduce ethical and legal challenges, including bias reinforcement, reduced transparency, privacy concerns, and diminished accountability. The complexity and opacity of modern AI models further exacerbate these risks, making it difficult to ensure fairness and compliance with ethical and legal standards.

Most AI systems today rely on Machine Learning algorithms, and the need for ethics and trust in AI has been emphasized through various regulations and guidelines. Frameworks such as the EU’s GDPR mandate the right to "meaningful explanations" of automated decision-making processes, while the AI Act advocates for explainability, transparency, and accountability in AI-driven decision-making.

Despite these efforts, the challenge of developing AI systems that are truly explainable, trustworthy, and compliant with regulatory frameworks remains unresolved. Many methodologies address aspects such as explainability, fairness, and accountability, but a comprehensive and effective framework is still lacking. Addressing these challenges requires collaboration across disciplines, including computer science, law, sociology, and ethics.

XKDD and Beyond is dedicated to advancing research on explainable, transparent, ethical, and fair AI-driven decision systems. This year, the workshop expands its scope to include the emerging topic of unlearning and its intersection with explainable AI (XAI). Unlearning, or the removal of specific knowledge from AI models, is a crucial challenge, especially in contexts requiring compliance with the "right to be forgotten" under GDPR. However, achieving effective unlearning is complex, as model parameters encode learned information in intricate ways.

XAI techniques play a vital role in operationalizing unlearning by providing insights into how decisions are made, identifying the influence of specific data points, and guiding targeted interventions to remove unwanted knowledge while preserving model performance. By integrating explainability with unlearning, AI systems can become more adaptable, accountable, and aligned with ethical and legal standards.

XKDD 2025 invites researchers and practitioners from academia and industry to explore the latest advancements in explainability, trust, and unlearning in AI. Join us in shaping the future of ethical and transparent AI-driven decision-making.

Topics of interest include, but are not limited to:

• XAI and Unlearning
• XAI for Trustworthy AI
• XAI for Social AI
• XAI to Align AI with Human Values
• XAI for Outlier and Anomaly Detection
• XAI methodologies for tabular data, images, text, and time series
• XAI for ethical, fair, and transparent AI systems
• XAI techniques for enabling unlearning in machine learning models
• The role of XAI in privacy preservation and regulatory compliance
• Case studies on XAI and unlearning applications in real-world scenarios
• Quantitative and qualitative evaluations of XAI
• Quantitative and qualitative evaluations of unlearning approaches
• XAI for Federated Learning
• XAI for Graph-based Approaches
• XAI for Visualization
• Interpretable Machine Learning
• Transparent Data Mining
• XAI for Fairness Checking
• Multi-level XAI
• Explanation, Accountability, and Liability from an Ethical and Legal Perspective

The call for paper can be dowloaded here.