Sweden's Government Offices ban staff from using ChatGPT and Claude
IA en un minuto newsroom · Editor: Jon Elgezabal
In 30 seconds
On Monday, October 26, a block takes effect: staff at the Swedish government's offices will no longer be able to open ChatGPT or Claude from their work computers. The move steers employees toward approved tools such as Microsoft Copilot, although some will keep access through an in-house platform. Ekot says the Tax Agency and SiS already apply similar blocks.
Sweden's Government Offices (Regeringskansliet) will block access to ChatGPT and Claude from staff work computers starting Monday, October 26, Ekot, the news programme of public broadcaster Sveriges Radio, reported on Friday.
The goal is for staff to use the tools the Government Offices have approved, Microsoft Copilot among them. Digitalization Director Mikael Wettercrantz declined an interview with Ekot but warned, in a comment posted on the Government Offices' intranet, that turning to AI services without an agreement may have "consequences for our information security."
The press office says some staff will keep access to ChatGPT and Claude through an AI assistant platform built by the Government Offices themselves. According to internal information seen by Ekot, Copilot's security has also been tightened after a review; the press office declined to explain what that change involves, saying it concerns security matters. The Government Offices had already limited the use of DeepSeek.
They are not alone. According to Ekot's survey, the Swedish Tax Agency (Skatteverket) and the National Board of Institutional Care (SiS) have already brought in similar blocks, and the Swedish Enforcement Authority (Kronofogden) is preparing a decision on the matter.
Ekot notes that the risks of using AI without rules have already led other agencies to act. One is the Swedish Transport Administration (Trafikverket), whose internal audit found this year a widespread and risky use of roughly ten different AI services, with examples involving both how sensitive the information entered was and how the generated answers were handled.
Why it matters · analysis and opinion
The block is not aimed at AI itself but at using it without a contract: the Swedish government keeps Copilot, which it has approved, and an in-house route for those who need ChatGPT or Claude. In the public sector, the question is no longer whether AI is used, but under what agreement and where the data fed into it ends up. The fact that the Tax Agency and SiS have done the same, and that Kronofogden is looking into it, points to a trend rather than a one-off case. For any organization that handles sensitive information, the practical step is to know which tools its staff really use and to offer an approved alternative before cutting access: the Trafikverket audit shows what happens without that control, with sensitive data entered into open services.
Source: Sveriges Radio · Written with the help of AI: how we make the news
