What you need to know about Cowork in Microsoft Copilot
As Microsoft continues to expand what Copilot can do, one area that’s starting to get more attention is “Co-work”, powered by Claude from Anthropic.
We’re seeing more clients ask the same questions:
- Is it enabled by default?
- Do my users have access to use it?
- Where is the data processed?
- And how does it fit within Microsoft’s security model?
Here’s what you need to know…
What is Co-work?
Co-work is designed to move Copilot beyond simply responding to prompts and into something that feels much closer to a digital teammate.
Instead of just asking a question and getting a one-off answer, Co-work allows you to work alongside AI in a more continuous, collaborative way. It becomes easier to delegate pieces of work, whether that’s drafting content, pulling together insights, refining ideas, or progressing tasks without having to start from scratch each time.
It’s a shift in how we work with AI, but an important one. You’re not just using a tool anymore; you’re working with something that can actively support how you think, plan, and operate day to day, helping you move from individual prompts to more meaningful outcomes.
Co-work is currently part of Microsoft’s Frontier programme with no confirmed date of when it will be generally available yet.
Is Claude (Co-work) enabled by default?
No, not in the UK and EU atleast. Unlike the traditional Copilot experience, which is powered by ChatGPT, Co-work runs on Claude through Microsoft’s new partnership with Anthropic.
Because of that, it is turned off by default in UK and EU tenants and needs to be enabled by an administrator at a tenant-wide level.
The service involves data processing in the United States, so Microsoft requires organisations to make a conscious, business-level decision before enabling it.
This isn’t something that quietly appears in the background. It’s something IT and the wider business should review and approve.
What happens to your data?
We understand that hearing your data is processed outside the EU can raise some initial concerns, which is exactly why this is positioned as a business decision.
However, it’s important to be clear on how data is handled when using Claude within Copilot:
- Your data is processed, not stored, by Anthropic
- Anthropic does not take ownership of your data
- Your data remains within your Microsoft 365 security and compliance boundary
In practical terms, that means all the controls you already rely on in Microsoft 365 Copilot, permissions, compliance policies, data protection, still apply.
Microsoft also formally recognises Anthropic as a subprocessor for its online services, which you can read more about here: Anthropic as a subprocessor for Microsoft Online Services
So, while the processing location is different, the governance model remains consistent with the rest of your Microsoft environment.
The key takeaway
Co-work with Claude is a powerful addition to Copilot, but it’s not something that should be enabled without thought.
It introduces:
- A different processing location (US)
- A requirement for admin-level enablement
- A need for clear business approval
But at the same time, it still operates within the Microsoft security, compliance, and governance framework you already trust.
New Copilot features are landing fast… are you ready for them?
It’s not just about what’s available, it’s about knowing what to enable, when, and why. We help organisations move beyond experimentation by identifying the right use cases, building a clear adoption strategy, and making sure the right governance and security foundations are in place.
