Sweden’s Government Offices, known as Regeringskansliet, will block direct access to OpenAI’s ChatGPT and Anthropic’s Claude on employees’ work computers starting October 26, 2026. The measure was reported by Swedish public broadcaster Sveriges Radio.
The restriction is intended to ensure that employees use AI services approved by the organization. Microsoft Copilot will remain one of the main alternatives available to staff.
Why is Sweden introducing the restriction?
Mikael Wettercrantz, the Government Offices’ director of digitalization, said in an internal communication that using AI services without an agreement could have consequences for information security.
One key concern is that employees might enter confidential or work-related information into external AI platforms. Without appropriate contractual arrangements, organizations may have less control over how such information is processed, stored and protected.
Sveriges Radio also reported that an internal audit at the Swedish Transport Administration identified risks associated with employees using various publicly available AI services. The concerns included the sensitivity of information entered into the tools and how AI-generated responses were handled.
Which AI tools can employees use instead?
Government staff will be directed toward approved tools, including Microsoft Copilot. The Government Offices have also reviewed Copilot’s security and strengthened its protections, although officials have not disclosed the details of those changes.
The restriction does not mean that every employee will completely lose access to ChatGPT and Claude. According to Sveriges Radio, some staff will retain access to the models through a dedicated AI assistant platform developed by the Government Offices themselves.
Other Swedish public authorities have introduced similar measures. These include the Swedish Tax Agency, Skatteverket, and the National Board of Institutional Care, known as SiS. The Swedish Enforcement Authority, Kronofogden, is also considering the issue.
What does this mean for AI adoption?
The decision highlights a growing challenge for organizations adopting generative AI: how to benefit from these tools while maintaining control over sensitive information.
Blocking access to particular services does not necessarily mean abandoning AI. Instead, organizations can establish approved platforms, contractual safeguards and clear rules governing what employees may enter into AI systems.
For government agencies, where employees may handle sensitive public-sector information, these safeguards are particularly important. The Swedish case illustrates that AI adoption increasingly involves not only choosing capable models, but also deciding how they can be used safely within an organization.
