From 26 October 2026, Sweden's Government Offices, the body that supports the prime minister and the cabinet, blocked ChatGPT and Claude on staff computers. According to Decrypt, which cites Swedish public radio, Microsoft Copilot is now the only approved AI tool for staff there. The decision says less about which AI is best and more about which tools have a formal agreement covering what happens to the data people type in.
What the government said
Mikael Wettercrantz, the Government Offices' director of digitalisation, told staff on the internal intranet that using AI services without a formal contract can have consequences for information security. He said the office had seen several examples of problematic use of open AI services, involving both the sensitivity of what people entered and how AI-generated answers were handled.
The underlying worry is simple. When someone uses a consumer AI account, what they type goes to the provider's servers under consumer terms. A business agreement sets out what the provider may and may not do with that data. Decrypt reports that the Government Offices reviewed Copilot and tightened its security settings afterwards, which it hadn't done for ChatGPT or Claude.
It isn't quite a ban on the models
The line between banned and approved is blurrier than the headline suggests. Decrypt reports that some staff can still reach both ChatGPT and Claude through an AI platform the office built internally, though it hasn't said how many people or under what limits. Microsoft also added Claude models to Copilot earlier in 2026. In the EU those models are switched off by default and an administrator has to turn them on, and the Government Offices hasn't said whether it has done so.
Other Swedish agencies are moving the same way
Decrypt reports that Sweden's Tax Agency and the National Board of Institutional Care already have similar bans on consumer AI tools, and that the Enforcement Authority is preparing its own decision. An internal audit at the Transport Administration found roughly ten AI services in use in what it called a widespread and risky way. Outside Sweden, the European Parliament switched off AI features on its work devices in February.
About a decade ago the same pattern played out with personal Dropbox and Gmail accounts. Staff found useful tools faster than IT could approve them, and most organisations responded with blocks first and business accounts later. AI tools are following the same path, only faster.
Four questions to ask your IT team this week
Which AI tools are approved for work, and on which accounts? Is there a business agreement in place, and does it stop our data being used for training? Which kinds of information are we allowed to put into them, for example client names, contracts or financial figures? And who do I ask if I want a new tool approved?
Once you have the answers, move anything confidential onto the approved tool and keep personal accounts for personal use. If your company hasn't approved anything yet, the Compare page sets out the business plans side by side, which makes a useful starting point for that conversation.