Sweden’s Government Offices block Claude and ChatGPT

From Monday, October 26, staff at Sweden’s Government Offices (Regeringskansliet) can no longer reach ChatGPT or Claude from their work computers. The block is meant to steer employees toward the AI services Regeringskansliet has approved, Microsoft’s Copilot among them, Sveriges Radio’s Ekot reported on October 23.

Digitalization director Mikael Wettercrantz declined an interview with Ekot. In a comment on the Regeringskansliet intranet, he wrote that using AI services without an agreement can have “consequences for our information security.”

The block has a carve-out. The press office says some staff will keep access to a dedicated platform for AI assistants that Regeringskansliet developed itself, and that platform still gives them ChatGPT and Claude.

Copilot is changing as well. After a review, Regeringskansliet raised the security of the approved tool, according to internal information Ekot has seen. The press office won’t go into detail about the tightening, citing security reasons.

This isn’t the first restriction. Regeringskansliet had already limited Deepseek, and, like the Norwegian Storting and the Danish Folketing, it has restricted both Deepseek and TikTok, SvD reported. Sweden’s parliament has taken a different route there. Johan Åhrman, the Riksdag’s information security officer, said bans on individual apps are not normally issued in the Riksdag and the focus is on education: staff are taught the handling rules for what information can be given to services such as TikTok or Deepseek, while the Riksdag follows developments and continuously analyses which types of services emerge and become popular. That leaves Deepseek off-limits in the Government Offices but free to use in the Riksdag.

Ekot’s survey shows Regeringskansliet isn’t the only one blocking these services. The Swedish Tax Agency (Skatteverket) and the National Board of Institutional Care (SiS) already have blocks in place, and the Swedish Enforcement Authority (Kronofogden) is preparing the question ahead of a decision.

Other agencies have run into the same risks. Earlier this year, an internal audit at the Swedish Transport Administration (Trafikverket) found widespread, risky use of about ten different AI services. The auditors wrote:

“We have taken note of several examples of problematic use of open AI services. The examples concern both the sensitivity of the information that was entered and how the AI-generated answers were handled.”