Sweden Limits ChatGPT and Claude Use in Government Offices

The Swedish Government Offices have restricted access to the browser-based versions of OpenAI’s ChatGPT and Anthropic’s Claude on work computers. The decision follows concerns that public employees may enter sensitive government information into AI services without sufficient contractual and technical safeguards.

The measure was prompted by an earlier review at the Swedish Transport Administration, which identified problematic uses of generative AI in government work. Authorities were particularly concerned that data submitted to external AI tools could be retained or used for model training where no appropriate data-processing and security arrangements are in place.

Sweden’s approach reflects a wider public-sector concern about so-called shadow IT. Mikael Wettercrantz, Director of Digitalisation at the Government Offices, warned that unapproved platforms can create information-security risks. The assessment is especially relevant where personal data, confidential documents, protected information, or security-sensitive material may be processed.

Microsoft Copilot has been recommended as an alternative after completing a security review and being covered by an agreement between Microsoft and the Swedish government. A limited group of officials may continue using ChatGPT and Claude through a dedicated public-authority platform for AI assistants, subject to tighter controls.

Similar restrictions have reportedly been introduced at the Swedish Tax Agency and the National Board of Institutional Care, while the Swedish Enforcement Authority is reviewing its position. ChatGPT and Claude now join DeepSeek among AI tools not approved for ordinary government use due to data-security concerns.