Official ONS data shows AI adoption in UK business has almost tripled since 2023, yet depth has barely moved. Most firms run one or two narrow use cases. Here is why breadth flatters your board, and why the real gap is a people problem, not a technology one.
Apple Intelligence has been cleared for China, where it will run on Alibaba and Baidu models rather than Apple’s own. One brand, two systems. That split turns a product detail into a documentation problem for anyone deploying AI across borders.
A special EU expert panel wants to flip the burden of proof for online child safety. Providers would have to show a service is safe by design before children under 13 could use it, and the same design features the Commission just cited against Meta are in scope.
The European Commission has opened a targeted consultation on data sovereignty, running to 8 September. It asks who can reach European data and under whose law. The policy outcome is uncertain. The internal exposure map an honest answer requires is valuable whatever Brussels decides.
A single US state has set a new bar for frontier AI safety, with recurring independent audits of the largest developers. European organisations are not bound by it. But the evidence it forces into the open changes what governance teams can now ask of their AI vendors.
Brussels softened the AI literacy duty in the Digital Omnibus, from ensuring a sufficient level to supporting its development. But it was not delayed, it has applied since February 2025, enforcement begins this August, and the EU’s skills agenda is turning literacy from a compliance checkbox into a competitiveness lever.
