New study maps EU, US, China AI rules, exposes compliance gaps

AI governance is shifting from voluntary ethics to enforceable, risk-based regulation, but the global landscape remains fragmented. A new arXiv paper (2608.14562) from Aasish Kumar Sharma and colleagues at the University of Göttingen compares the EU, US, and China regulatory frameworks across four dimensions: risk classification triggers, binding obligations, enforcement mechanisms, and operationalization of FAIR principles. The authors stress-test their comparative matrix on three high-impact domains: EEG-guided rehabilitation robotics, AI-enabled debt collection in future CBDC ecosystems, and AI-driven allocation of scarce GPU resources in emerging AI Factory infrastructures. The paper, accepted at the 50th IEEE COMPSAC 2026 in Madrid, highlights that while all three regimes claim risk-based governance, they diverge sharply on what triggers oversight and how compliance is verified.

Across these domains, the researchers identify three recurring gaps: weak interoperability mandates between regimes, difficult operationalization of overlapping obligations (AI law + sector regulation + data protection), and under-specified governance for critical digital infrastructure. To address this, they introduce Knowledge Blocks—a machine-checkable compliance artifact pattern built on RDF/OWL, SHACL, and PROV-O. This allows organizations to encode regulatory requirements as semantic graphs, validate compliance with SHACL rules, and track provenance with PROV-O, enabling audit-ready compliance-by-design across multiple jurisdictions. The authors argue this approach reduces the burden of manually mapping obligations and provides a foundation for automated regulatory checks, especially for high-stakes AI systems operating internationally.