Source: Thomson Reuters

Thomson Reuters discusses the growing concern around data fragmentation and its impact on the performance, reliability, and legal compliance of artificial intelligence systems. Fragmented data refers to datasets spread across disconnected systems or collected inconsistently, leading to risks in bias, error propagation, and noncompliance with data protection laws.

The article outlines how data fragmentation can hinder the interpretability of AI outputs and reduce effectiveness in areas like legal research, fraud detection, and compliance tools. Moreover, incomplete or inaccessible data complicates risk assessments and audit trails required under regulatory regimes like GDPR and the AI Act.

Experts suggest solutions such as improved data governance, centralized data models, and AI systems designed with interoperability and legal traceability in mind.