Employees within the Swedish government have been prohibited from accessing browser-based versions of OpenAI’s ChatGPT and Anthropic’s Claude on work-issued computers. This directive follows security concerns regarding the potential for sensitive government data to be processed by these models without proper safeguards.
The decision was prompted by a review conducted earlier this year at the Swedish Transport Administration, which identified instances of what officials described as "problematic" usage of AI platforms in the execution of official government duties. Authorities expressed apprehension that inputting sensitive information into these models, in the absence of a formal agreement governing data protection and security, could result in the data being utilized for model training and storage.
Mikael Wettercrantz, the director of digitalisation at the Government Offices of Sweden, noted that the utilization of platforms lacking regulated security agreements—often referred to as shadow IT—poses risks to the state's information security. As a secure alternative, the government is recommending the use of Microsoft’s Copilot, which has successfully completed a security screening and is subject to a formal agreement between the provider and the Swedish state.
While government staff are still permitted to utilize AI as a tool for their tasks, they must now adhere to more stringent security protocols. Only a limited number of employees will be granted continued access to ChatGPT and Claude, and such access will be restricted to a specialized platform for AI assistants built by Swedish authorities. Similar restrictions have been implemented at the Swedish Tax Agency and the National Board of Institutional Care, according to reports from the public broadcaster Sveriges Radio, while the Swedish Enforcement Authority is currently conducting its own review of the matter.
These platforms join DeepSeek on a growing list of AI tools currently unapproved by the Swedish government due to data security fears. This development follows a recent announcement from the United States, which banned all products and services originating from the Chinese AI company DeepSeek.
