Sweden's Government Offices, which include the prime minister's office and the ministries, have since October 26 blocked staff access to OpenAI's ChatGPT and Anthropic's Claude on work computers. According to reports, the block covers the browser versions of both services, and staff are directed to Microsoft Copilot.
Why now
According to the Government Offices' digitalization director, Mikael Wettercrantz, using AI services without a proper contract can harm information security. According to reports, he also cited "several examples of problematic use of open AI services". Copilot, by contrast, has been reviewed by the government, and its security settings adjusted for this use. According to one report, both models remain available to staff through an approved Swedish platform.
Not a leak, but caution
To be clear: the reports do not claim that Swedish government data leaked or was used to train the models. This is a pattern of risky use, not a proven breach. Nor is it a national ban, but a policy for government staff computers.
The trend
Sweden joins a series of governments and organizations trying to control how employees use AI. The problem is familiar in Israel too: employees paste sensitive documents into chatbots to save time, without knowing where the data goes. Sweden's solution, one approved tool instead of a blanket ban, is probably the direction many organizations will take.
