The key facts

  • Swedish civil servants are banned from using the browser versions of ChatGPT and Claude on work computers
  • The reason is the risk of confidential data ending up in model training datasets without data protection agreements
  • Microsoft Copilot has been recommended as an alternative; restrictions have also been introduced at the Tax Agency and the National Board of Institutional Care

Ban for civil servants

Employees of Swedish government agencies can no longer use the browser versions of OpenAI's ChatGPT and Anthropic's Claude on their work computers. According to Euronews, the decision is linked to concerns about the handling of confidential data. Restrictions have also been introduced at the Swedish Tax Agency and the National Board of Institutional Care, while the Swedish Enforcement Authority, according to Sveriges Radio, is considering a similar move.

The reason: 'shadow IT' and training data

A review of AI tools used by staff at the Swedish Transport Administration earlier this year revealed 'problematic' use of these platforms in government work. Officials fear that entering confidential information into models without a data protection agreement could result in that information ending up in training datasets. Mikael Wetterkrantz, Chief Digital Officer at the Swedish Government Offices, said that using platforms not covered by such agreements — so-called 'shadow IT' — 'can have consequences for our information security'.

Alternatives and new rules

Microsoft Copilot has been recommended as a replacement; it has passed a security review and is covered by an agreement between the company and the Swedish government. Civil servants will retain the ability to use AI in their work, but under stricter control: only a limited number of employees will be able to continue working with ChatGPT and Claude — and only through a special platform for AI assistants developed by the Swedish authorities. ChatGPT and Claude have joined DeepSeek on the list of AI services not approved by the government; last week, the United States announced a ban on all products and services of the Chinese company DeepSeek.

Why it matters

The decision affects thousands of civil servants whose daily access to popular AI tools is now restricted, and it sets a practical standard: work with confidential data is possible only through services covered by data protection agreements, such as Copilot. For other agencies, including the Enforcement Authority, the matter is still under consideration, so the list of affected institutions may grow.