
Sweden's Government Offices Block ChatGPT and Claude, Keep Microsoft Copilot
Sweden's Government Offices will block ChatGPT and Claude on staff computers from 26 October 2026 and keep Microsoft Copilot as the only approved AI tool, after a security review. Director of digitalisation Mikael Wettercrantz cites sensitive data entered into open services and tools used without a formal contract. The Tax Agency and the National Board of Institutional Care have similar bans. The article notes Copilot can include Claude models, and the Government Offices has not said whether that is enabled.
What it claims
From 26 October 2026, Sweden's Government Offices, the body supporting the prime minister and cabinet, will block ChatGPT and Claude on staff computers. Microsoft Copilot is the only approved tool. Mikael Wettercrantz, director of digitalisation, posted the reasoning on the staff intranet. He cited "several examples of problematic use of open AI services," covering both the sensitivity of what staff entered and how they handled the AI-generated answers. Sveriges Radio first reported the decision.
How it works
The line is drawn by contract and review, not by vendor or model. Copilot passed because the Government Offices ran a dedicated security review of it and then raised its security settings. It did neither for ChatGPT or Claude. Decrypt gives two reasons a formal agreement matters. First, consumer tools without one send queries to external company servers, and U.S. law lets authorities compel those companies to hand over data. Second, an agreement defines what the provider may do with employee inputs.
The ban has an exception. Some employees can still reach both blocked tools through an AI platform the Government Offices built internally. The press office has not said how many staff this covers or what restrictions apply.
Evidence
Other agencies are moving the same way:
- The Tax Agency and the National Board of Institutional Care already ban consumer AI tools.
- The Enforcement Authority, which handles debt collection and court orders, is preparing its own decision.
- An internal audit at the Transport Administration found about 10 AI services in use there in a "widespread and risky way."
- In February, the European Parliament went further. It disabled AI features on all work devices, Copilot included, because it could not guarantee the security of the data those tools processed.
Limits
The "banned" and "approved" categories may overlap. Microsoft added Claude models to Copilot in early 2026. In the EU, those models are off by default, and an administrator must deliberately enable them. The Government Offices has not confirmed whether its Copilot instance has Claude switched on.
The article does not describe the security settings changed in the Copilot review. It also gives no detail on the internal platform's architecture or data handling.
Organisations named here
- Sweden's Government Offices
- Sveriges Radio
- Microsoft
- OpenAI
- Anthropic
- Swedish Tax Agency
- National Board of Institutional Care
- Swedish Enforcement Authority
- Swedish Transport Administration
- European Parliament
- Mikael Wettercrantz
- Transport Administration
Technologies named here
- Microsoft Copilot
- ChatGPT
- Internal AI platform of the Government Offices
- Claude
Related in ai-policy


Future of Life Institute AI Safety Index — Summer 2026

Lessons from the hacks
