Microsoft turns on Claude by default in UK Copilot, The Register reports

Microsoft will switch on Anthropic’s Claude models by default for UK Copilot customers from 9 December, The Register reported. Anthropic processes that data in the US.

Richard Speed wrote for The Register that UK organisations with no setting chosen by then will have Claude enabled. Administrators can still turn it off. Leaving the setting untouched means opting in.

Anthropic works as a Microsoft subprocessor, a third party that handles customer data under Microsoft’s contracts and oversight, according to Microsoft’s Copilot documentation. Claude is already on by default for most of Microsoft’s commercial customers. The EU, the European Free Trade Association and the UK have so far been the exceptions.

“This affirms Microsoft’s commitment to offering choice between leading AI models while maintaining enterprise-grade security and compliance,” Microsoft wrote.

Outside the EU Data Boundary

The same documentation says Claude models in Copilot fall outside Microsoft’s EU Data Boundary. They are also excluded from in-country processing commitments. The UK is not covered by the EU Data Boundary, so Microsoft is changing the default there, according to The Register.

The change covers Claude in Copilot, Researcher, Copilot Studio, Power Platform and Copilot in Microsoft 365 apps. Microsoft is also retiring a setting that limited Claude to Microsoft 365 apps.

Some advanced Anthropic models, such as Claude Fable 5 and Claude Mythos 5, stay off by default everywhere. Anthropic keeps data from those models, mostly for up to 30 days, so administrators must opt in to them separately, Microsoft said.

The same week, Anthropic launched Claude Sheets and other office tools that compete with Microsoft’s own.

Whitehall’s US tech bill

The UK public sector’s dependence on US technology is the subject of a new report from the Institute for Public Policy Research (IPPR). Less than £1 in every £6 spent with the 150 largest technology suppliers stays with British companies, The Register reported on Wednesday.

About 42.5% of that spending appears to go to UK companies. The share falls to 15.4% once the researchers looked past headquarters to ownership, staff and resellers. The US gets the most, with £9.2bn of the £19.2bn spent with the largest suppliers.

The report, called Locked In, found that AI often reaches government through existing cloud and software deals. NHS England’s Copilot rollout reached 505,000 staff. It ran under a national licensing deal signed in June 2023, before Copilot was generally available.

IPPR also pointed to US restrictions on Anthropic models. It cited limits on early access for the UK’s AI Security Institute too.

The government has committed £500m to a Sovereign AI Fund. IPPR wants Whitehall to buy technology as a single customer and make it easier to switch suppliers. It also wants the UK to work with France, Germany, Japan, and South Korea on access to advanced AI.

“But we cannot do it alone, and building domestic capability can’t be our only bet,” said Roa Powell, senior research fellow at IPPR.

UK administrators who want to keep Claude switched off in Copilot have until 9 December to change the setting.