Starting next week, the Swedish Government Offices will restrict employees from accessing ChatGPT and Claude directly via work computers. Staff members are instructed to transition to approved AI tools, specifically Microsoft Copilot, which has undergone internal security reviews and enhanced protection measures.
Mikael Wettercrantz, the Chief Digital Officer of the Government Offices, stated that using unauthorized AI services poses significant risks to information security. While specific security protocols for Copilot remain undisclosed, the administration emphasizes the necessity of managing AI usage to prevent data leaks.
Other government bodies are implementing similar restrictions. The Swedish Tax Agency and the National Board of Institutional Care have already blocked access to unauthorized platforms. Meanwhile, the Swedish Enforcement Authority is currently evaluating whether to adopt comparable limitations.
The Swedish Transport Administration previously identified risks associated with unregulated AI usage. An internal audit revealed that employees were utilizing approximately 10 different AI services, leading to concerns regarding the sensitivity of input data and the reliability of AI-generated responses.
Despite these broad restrictions, some government employees may still access ChatGPT and Claude through internally developed AI assistant platforms. This approach allows the government to maintain control over data processing while still providing staff with access to advanced language models.
