The European Commission and Japan's Ministry of Internal Affairs and Communications announced expanded cooperation on AI governance on 5 May 2026, including commitments to align regulatory approaches and share information on AI system failures [source].

The joint statement follows the EU-Japan Digital Partnership dialogue and establishes working groups on AI safety, testing methodologies, and incident reporting frameworks. Both jurisdictions committed to quarterly exchanges on observed model degradation and behavioural anomalies in deployed systems.

Under the arrangement, the EU AI Office will share technical assessments of general-purpose AI models with Japanese counterparts, while Japan will provide data from its AI Safety Institute on model performance drift. The agreement does not create binding obligations but establishes information-sharing protocols that could inform future regulatory action in both regions.

The announcement references ongoing challenges with AI system reliability, noting that both the EU and Japan have observed instances where models produce outputs inconsistent with provider documentation. The working groups will examine whether current testing frameworks adequately capture post-deployment performance changes.

The cooperation extends beyond AI to include quantum computing standards and semiconductor supply chain resilience, but the AI governance component represents the most detailed commitment in the statement. Both parties indicated they would invite other jurisdictions to participate in technical working groups on a case-by-case basis.

The EU AI Office has not specified which providers or models will be subject to the enhanced information sharing, stating only that the arrangement covers systems meeting the general-purpose AI definition under the EU AI Act. Japan's participation follows its establishment of domestic AI safety evaluation capabilities in early 2026.