Up Close
AI turns a software shortcut into overhead, while privacy law turns an AI safeguard into a stoppage
Bremer Bäder GmbH switched off Lynxight’s AI drowning-detection camera system at Bremen’s Westbad after the state data protection authority found that the required GDPR data-protection impact assessment had not been carried out, and the system is to remain offline until that review is completed.
This suggests that the practical barrier for AI in public-facing settings is not only whether the model works, but whether the operator can document the risks before it goes live. For Microsoft, the same pressure is likely to show up less as a ban than as a buying condition: if AI is sold into schools, hospitals, offices or cities, governance paperwork stops being a side task and becomes part of the product. That also reaches beyond Microsoft, because any supplier of AI cameras, assistants or agents into regulated environments appears to need deployable controls and audit trails, not just a capability claim.
Shopify said it is abandoning React Native and moving its mobile app development back to native implementations because AI-assisted coding tools now make native development efficient enough to replace the cross-platform compromise.
We read this as a shift in what AI changes inside software teams: it does not only add features, it can remove the old reason for a shared abstraction layer. For Microsoft, that is likely to matter wherever Copilot and related coding tools are sold as productivity gains, because the gain becomes more concrete when teams can keep platform-specific code without paying the former labour penalty. The wider effect appears to be that AI coding support strengthens local optimisation over lowest-common-denominator software stacks, which changes tool choices for companies far beyond Shopify.