Microsoft makes major AI U-turn following user revolt — will let Teams users turn off Copilot, Facilitator and Recap
THE SO WHAT
Forced AI in core workflows is hitting user resistance hard enough to trigger reversals — opt-out and control are now table stakes for enterprise adoption. If you’re rolling out assistants internally, treat consent, transparency, and easy disablement as product requirements, not change-management afterthoughts.
READ THE SOURCE
MORE FROM THE WIRE
Show HN: Scan your AI agents for dangerous capabilities
Treating agent capability scanning as a discrete tool is an early signal that “red-teaming your own agents” will become a standard control, not a niche research task. If you’re piloting agentic workflows, add automated capability checks to your pre-production checklist before you scale usage.
Applied AIAlibaba is banning its workers from using Claude Code as US v China AI battle heats up
Alibaba blocking Claude Code over data tracking and geopolitics shows that AI dev tools are now entangled with national security and data-sovereignty policy. Multinationals need a country-by-country matrix for which AI services are allowed, plus clear internal alternatives when a tool gets geo-blocked overnight.
Applied AIImport AI 464: Fables writes GPU kernels; AI automation; and analog computation
If models can reliably write GPU kernels, the bottleneck in high-performance AI work shifts from low-level optimization to problem framing and verification. Teams doing custom acceleration should start experimenting with AI-assisted kernel generation now, but keep a tight human review loop around performance and correctness.
Applied AIA senior FCA official says Britain should weigh regulating AI models directly
If the FCA moves toward direct oversight of models like ChatGPT, Claude, and Gemini, financial services firms will inherit model-level compliance obligations, not just application-level ones. UK-regulated institutions should start logging which foundation models touch customer decisions and prepare for audit-style questions about their behavior and provenance.