Reuters reported that the US Commerce Department removed details of a previously announced arrangement for Microsoft, Google and xAI to give government scientists early model access for security testing. The removed material concerned checks for risks including cyber misuse and military applications before public deployment. 

The Guardian reported that the US tech delegation’s China visit is unfolding alongside renewed debate over AI model review and national-security controls. The coverage linked commercial AI diplomacy with pressure for stronger pre-release oversight of advanced systems. 

Regulation

  • BaFin said it will use targeted ‘IT spotlight’ inspections as AI increases cyber and operational risk in the financial sector. President Mark Branson described faster supervisory checks focused on IT resilience rather than broad full-scope reviews. 

  • NIST NCCoE held the third working session on usability of its Cyber AI Profile. The session sought input on how organisations can adopt AI while prioritising cybersecurity risks linked to AI development and deployment. 

Academia

  • ScienceDirect published research on Article 50(1) of the EU AI Act and transparency duties for AI systems intended for direct interaction with individuals. The article examines how disclosure rules can work in practice when users interact with AI rather than merely receive AI-assisted outputs. 

Events

  • Erasmus University Rotterdam will host the Law, AI and Regulation Conference on 11 and 12 June 2026. The programme brings together scholars, practitioners and policymakers on responsible AI governance, the EU AI Act and regulatory implementation. 

  • UNIDIR will hold the Global Conference on AI, Security and Ethics in Geneva and online on 18 and 19 June 2026. The event focuses on AI governance in security, defence, international law and cross-border coordination. 

Sources: Reuters, The Guardian, BaFin, NIST NCCoE, ScienceDirect, Erasmus University Rotterdam, UNIDIR

Tags
bafin
cyberai
article50
voicepilling
nccoe
modelrisk