By Roman Aglaganov
Sweden’s Government Offices, which run the work of the prime minister and the ministries, blocked the web versions of OpenAI’s ChatGPT and Anthropic’s Claude on staff computers from Monday 26 October. Staff are being directed to Microsoft’s Copilot instead. Sveriges Radio first reported the plan on 23 October, and the Government Offices confirmed it to the broadcaster SVT.
Why
The stated reason is information security. Mikael Wettercrantz, the Government Offices’ director of digitalisation, said that using AI services without a formal agreement can have “consequences for our information security”, Euronews reports. Officials worried that sensitive information typed into a consumer service, with no contract setting out how it is protected, could be stored or used for training. Notebookcheck, citing Decrypt, quotes him writing on the staff intranet: “We have seen several examples of problematic use of open AI services.”
Copilot is covered by a security agreement between Microsoft and the Swedish government, and the Government Offices says it reviewed the service and tightened its settings before approving it. As Notebookcheck points out, the difference is the contract and control over settings, not the underlying technology. The Government Offices has not said whether Anthropic’s models, which Microsoft also offers inside Copilot, are switched on.
Not a full ban
Some staff can still use ChatGPT and Claude through Intric, a Stockholm company whose platform for building AI assistants runs on servers in Sweden and which won a contract with the Government Offices in July. The block does not cover personal devices or the public. Other Swedish agencies have taken similar steps: the Swedish Tax Agency and the National Board of Institutional Care have restricted the services, and the Enforcement Authority is considering it.
Sweden had already limited the Chinese chatbot DeepSeek. The European Parliament switched off AI features on work devices in February for similar reasons, Notebookcheck notes. None of the reports we found included a response from OpenAI or Anthropic.
The lesson applies beyond governments. A consumer chatbot account runs on the provider’s standard terms, while a business contract can set where data is processed, how long it is kept and whether it may be used for training.
Disclosure: AI Pulse is written with AI tools, including Claude, which is made by Anthropic.
