Microsoft will switch on Anthropic's Claude artificial intelligence models for UK customers of Microsoft 365 Copilot on Dec. 9 unless their administrators have already chosen a setting, The Register reported Tuesday. Because Anthropic processes that data in the United States, UK organizations, and U.S. multinationals with UK tenants, face a cross-border data decision that will be made for them if they do nothing.
What Happened
According to The Register, UK Copilot tenants, the separate company environments in Microsoft's cloud, will have Claude enabled from Dec. 9 if admins have not picked an option, so silence counts as opting in. Microsoft is also retiring a narrower setting that limited Claude to Microsoft 365 apps.
Anthropic works inside Copilot as a Microsoft subprocessor, a supplier handling customer data under Microsoft's contracts, including its Data Protection Addendum. Microsoft's online documentation says Claude is already on by default for most commercial customers, while the European Union, the European Free Trade Association (EFTA) and the UK are currently off by default. Microsoft said in a statement quoted by The Register that the move "affirms Microsoft's commitment to offering choice between leading AI models."
Why It Matters
Microsoft's documentation says Claude models in Copilot, Researcher, Copilot Studio, Power Platform and Copilot in Microsoft 365 apps are excluded from the EU Data Boundary, its pledge to keep European customers' data in the region, and from in-country processing commitments. The Register said the boundary does not cover UK tenants, which it presented as the reason the UK default is changing now.
For boards and data-protection officers, the issue is that staff prompts and documents could start reaching a second AI provider in another country without a deliberate decision. A separate class Microsoft calls "Anthropic models with Data Retention," such as Claude Fable 5 and Claude Mythos 5, falls under Anthropic's own terms instead, and Anthropic may keep flagged content for up to two years. Those models stay off everywhere unless an admin opts in.
Claude is not offered to federal Government Community Cloud customers or to GCC High and Defense Department tenants, Microsoft says.
What to Watch
Before Dec. 9, an AI Administrator or Global Administrator can turn off Anthropic as a subprocessor in the Microsoft 365 admin center or limit access to chosen Microsoft Entra ID security groups, according to Microsoft. Organizations should check whether U.S. processing fits their data-protection obligations and customer contracts. Signals to watch include any comment from the UK Information Commissioner's Office and whether Microsoft later changes the EU and EFTA defaults.
