Sweden's government blocks ChatGPT and Claude, keeps Copilot

Sweden's Government Offices is blocking ChatGPT and Claude. From Monday, October 26, the browser versions of both chatbots will no longer be reachable on staff computers, as Sveriges Radio first reported. The Government Offices includes the Prime Minister's Office and the ministries. It confirmed the block in an email to broadcaster SVT. Staff are directed to Microsoft Copilot instead.

Copilot was reviewed first

The reason given is information security. According to Decrypt, Mikael Wettercrantz, director of digitalization at the Government Offices, wrote on the staff intranet that using AI services without a formal agreement can affect information security. "We have seen several examples of problematic use of open AI services," he wrote, as quoted by Decrypt. The Government Offices says it reviewed Copilot and then tightened the service's security. What separates Copilot is the contract and control over its settings, not the model. Staff can still reach OpenAI and Anthropic models through Intric, an internal platform, SVT reports.

The Government Offices has not said whether Anthropic's models, which Microsoft also offers in Copilot, are enabled. In the EU, they are off by default in Copilot and have to be turned on by administrators.

Other agencies already restrict chatbots

The Government Offices is not the first Swedish authority to take this step. According to SVT, the Swedish Tax Agency and the National Board of Institutional Care, which runs youth homes and addiction treatment facilities, already have similar blocks in place. The Enforcement Authority is preparing its own decision. An internal audit at the Transport Administration found around 10 AI services in widespread and risky use, Decrypt reports. Outside Sweden, the European Parliament disabled AI features on work devices in February because it could not guarantee data security.

What this means for your own account

The reasoning applies to private users too. With a regular account, your prompts are processed on the provider's servers under its standard terms. A business contract can set how long data is stored, where it is processed and whether it can be used for training. On consumer ChatGPT and Claude accounts, a setting in the privacy controls decides the training question. Our guide explains how to opt out of AI training at OpenAI, Google and Meta. A recent study also found that chat data from nine AI chatbots reached ad trackers.

I write about IT security and artificial intelligence. Cybersecurity is one of my regular topics, among others for Golem.de. At Notebookcheck, my focus is on practical AI for everyday users. I have worked hands-on with AI for years, most recently with a focus on AI agents and their use in everyday work, and I test tools and models myself rather than just reading about them.