Sweden tightens ai rules to protect government information
Sweden is restricting ChatGPT and Claude on government work computers over information security concerns, directing staff towards approved AI tools such as Microsoft Copilot.
Sweden is tightening controls on artificial intelligence in government workplaces, with several public agencies restricting access to unapproved AI services over concerns about information security.
From October 26, Sweden’s Government Offices will block ChatGPT and Claude on staff work computers and direct employees towards approved tools, including Microsoft Copilot. Digitalisation Director Mikael Wettercrantz warned that using AI services without an agreement could create risks for government information security, according to Swedish public broadcaster Sveriges Radio.
The restrictions extend beyond the Government Offices. Sweden’s Tax Agency and National Board of Institutional Care have also blocked access to the services, while the Enforcement Authority is considering its approach. Separately, an internal review led the Government Offices to strengthen security measures for Copilot, although officials have not disclosed the details.
The policy does not mean ChatGPT and Claude are banned for personal use across Sweden. Some government staff will also retain access to both services through a dedicated AI platform developed by the Government Offices.
The measures highlight a growing challenge for public institutions: allowing employees to benefit from AI while ensuring that sensitive work information is handled through approved systems. As government agencies expand their use of AI, security agreements and clear workplace rules are becoming an important part of adoption.
