Starting October 26, the Swedish Government Offices imposed restrictions on employees’ access to the AI-powered chatbots ChatGPT and Claude, citing information security risks.
Instead of these services, employees will be encouraged to use the approved Microsoft Copilot tool, although access to ChatGPT and Claude will remain possible through a special internal platform maintained by the agency.
Mikael Wetterkrantz, Head of Digital at the Government Offices, noted that the use of such services could have implications for information security, which is why security measures in Copilot have been strengthened.
The restrictions do not constitute a complete ban on OpenAI and Anthropic models, as employees can continue to work with these chatbots via the Intric platform, which provides access to AI assistants.
In addition to the Government Offices, other Swedish government agencies, such as the Tax Agency and the National Board of Institutional Care, have also implemented similar restrictions due to identified risks associated with the use of publicly available AI services.
