Sweden blocks ChatGPT and Claude on government computers but keeps Copilot
Sweden's Government Offices cut off two popular chatbots on staff machines over data security, showing how workplaces now approve AI tools one by one.
The short version
- ChatGPT and Claude are blocked on staff computers in Sweden's Government Offices.
- The rule covers the Prime Minister's Office and the government ministries, not the whole country.
- Swedish media say it took effect on October 26.
- Copilot remains allowed because it was reviewed, including its security settings.
- Officials cited sensitive information being typed in and how AI answers were handled.
Why it matters
Your own workplace may soon sort AI tools into approved and not approved, if it has not already. The Swedish case shows the question is less about whether AI is useful and more about where your words go when you type them in. Until you know your employer's rule, treat any chatbot as a place where private details should not be pasted.
What Sweden actually did
As Times Now on MSN reported, Sweden has stopped staff in the Government Offices from using ChatGPT and Claude on their work computers. Both chatbots come from US companies. The stated reason is information security.
The Government Offices is the body that includes the Prime Minister's Office and the government ministries. So this is a rule about one workplace, albeit an important one. It does not mean ChatGPT has been outlawed in Sweden, and ordinary people and businesses can still use it.
According to Swedish media reports, the restriction began on October 26. Employees are now expected to stick to AI tools that have gone through an internal check, and Microsoft's Copilot is on that list.
The worry: what staff type in
People who work in government handle material that is not meant to be public, such as drafts of policy, official documents and the day-to-day business of ministries. If someone pastes that kind of text into a chatbot, it leaves the building. Without proper safeguards, that creates a security risk.
Mikael Wettercrantz, the official in charge of digitalisation for the Government Offices, was reported to have cautioned staff that using AI services with no formal agreement in place could put information security at risk. In other words, the concern is not that the chatbots are bad at their job. It is that there was no contract or review covering how the information is handled.
The government also said it had seen cases of problematic use. These involved how sensitive the information put into the tools was, and how staff dealt with the answers the tools produced. That second point is easy to miss: the risk runs both ways, since a chatbot's output can be wrong or badly sourced and still end up in an official document.
Why Copilot got a pass
Copilot is Microsoft's AI assistant. It remains allowed because the Government Offices looked at the service and its security settings before letting employees use it. The report does not go into the technical details of that review, so we cannot say what specifically made the difference.
What the decision shows is a process, not a verdict on which chatbot is smarter. A tool is allowed once an organisation has checked it and is comfortable with the terms. Tools that have not been through that check are held back, however popular they are with the public.
That is why this counts as two blocked and one allowed rather than a ban on AI. The government is not telling staff to avoid the technology. It is telling them which versions they may use for official work.
Where your words go
The simple idea behind the move is that text typed into a chatbot usually travels to the company that runs it. At home, that may be a fair trade for a quick answer. At work, it can mean client names, contracts or internal plans leaving your organisation without anyone having agreed to it.
Chatbots can help people write documents, summarise material and finish tasks faster, which is why employees reach for them. The Swedish decision does not dispute that. It argues that the convenience has to come with security checks, and that a worker should not have to guess whether a tool is safe.
- Approved list: some workplaces now name the tools staff may use.
- Formal agreement: the Swedish warning points to having a proper arrangement with a provider.
- Case by case: each tool is judged on its own settings, not lumped together as "AI".
What to expect next
Other governments and large employers face the same problem: staff already use chatbots, often without asking. Sweden's step is a clear example of one answer, which is to block first and approve after review. Whether others copy it, or choose looser guidance, is still open.
The report does not say whether ChatGPT or Claude could be added to Sweden's approved list later. That would presumably depend on the kind of review and agreement the Government Offices mentioned. For now, the practical effect is limited to one set of office computers, but the logic applies far beyond Stockholm.
What to watch
- Will other governments or large employers adopt a similar block-then-approve approach?
- Could ChatGPT or Claude later be approved after a formal agreement?
- What did the Copilot security review actually check?
Try it
Ask your manager or IT team whether your workplace has an approved list of AI tools. If not, ask for one, and keep client names, contracts and private details out of chatbots until you hear back.
Sources and links
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.
Listen to the AI Wire headlines
Listen to the AI WireListen to our daily AI shows every morning: on your commute, in the shower, wherever you are
Listen to the daily showsFirst reported by Times Now on MSN · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.
