
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 leave Microsoft Copilot as the only approved AI tool, citing information security. Digitalisation director Mikael Wettercrantz said unvetted AI use had involved sensitive inputs and poorly handled outputs. The Tax Agency and National Board of Institutional Care have similar bans, and the European Parliament disabled AI features in February 2026. Whether Copilot has Claude models enabled is unconfirmed.
What it is
Sweden's Government Offices, the body supporting the prime minister and cabinet, will block ChatGPT and Claude on staff computers from Monday 26 October 2026. Microsoft Copilot stays as the only approved AI tool. Sveriges Radio reported the decision. Mikael Wettercrantz, director of digitalisation, announced it on the staff intranet.
The reasoning
Wettercrantz said using AI services without a formal contract can have "consequences for our information security." The stated problems were the sensitivity of what staff entered and how AI-generated answers were handled.
The article adds the underlying concern. Consumer AI tools without formal agreements send queries to external company servers, and U.S. law lets authorities compel those companies to hand over data. A formal security agreement, at minimum, defines what the provider may do with employees' inputs.
Copilot passed because the Government Offices reviewed it and raised its security settings afterwards. No such review was done for ChatGPT or Claude.
Scope and exceptions
- 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 people this covers or what restrictions apply.
- The Tax Agency and the National Board of Institutional Care already have similar bans on consumer AI tools.
- The Enforcement Authority is preparing its own decision.
- An internal audit at the Transport Administration found about 10 AI services in use in a "widespread and risky way."
- In February 2026 the European Parliament disabled AI features on all its work devices, Copilot included, because it could not guarantee the security of the data processed.
Caveats
Microsoft added Claude models to Copilot in early 2026. In the EU, Anthropic's models default to off and administrators must enable them. The Government Offices has not said whether its Copilot instance has Claude switched on, so the line between the banned and approved tools may be blurrier than the announcement suggests.
The article gives no details of the Copilot security settings that were changed, and no contract terms. The reported basis for the ban is the absence of a vetted agreement, not a stated flaw in either blocked product.
Organisations named here
- European Parliament
- Mikael Wettercrantz
- National Board of Institutional Care
- shadow AI
- Sveriges Radio
- Swedish Enforcement Authority
- Swedish Government Offices
- Swedish Tax Agency
Technologies named here
- ChatGPT
- Internal AI platform built by the Government Offices
- Claude
- Consumer AI services without formal agreements
- AI features on European Parliament work devices
- Microsoft Copilot
Related in GRC

Orca Security: NIST AI Risk Management Framework (AI RMF) Explained

FOSSA: CISA Releases the 2026 SBOM Minimum Elements
Teach Yourself to Phish | Huntress
