The explanation dialogues: an expert focus study to understand requirements towards explanations within the GDPR

Current XAI explanations fall short of legal requirements: Legal experts found that existing explainable AI explanations are "hard to understand and lack information," suggesting that many current XAI methods may not adequately meet the transparency obligations required under the GDPR.

Conflicting interests between data controllers and individuals create compliance challenges: The study identified that "different interests of the data controller and subject" can arise from explanations, highlighting a fundamental tension in how AI systems are explained—what benefits the company providing the AI may not serve the person whose data is being used.

Explanation design matters for legal compliance: The researchers recommend that XAI developers pay careful attention to "the presentation, choice, and content of an explanation," indicating that how explanations are formatted and what information is included are not just technical choices but have legal implications under GDPR.

Multiple legal gray areas remain unresolved: The study identifies several areas requiring further legal discussion, including "the contestability of explanations, transparency thresholds, intellectual property rights as well as the relationship between involved parties," suggesting that current regulations and XAI practices have not yet fully addressed how explanations should function in practice.

End-user considerations are critical but often overlooked: Recommendations specifically address "the end-user," emphasizing that explanations must be designed with the actual person receiving them in mind, not just to satisfy regulatory checkboxes.