Sweden Blocks ChatGPT and Claude on Public Sector Computers
Public sector employees across the Government Offices of Sweden are no longer permitted to access the web versions of ChatGPT and Claude on government-issued computers following data security restrictions that took effect on Monday, October 26, 2026.
The restrictions target the direct use of unvetted artificial intelligence platforms operating without formal, approved data protection agreements. Public officials cited significant risks that sensitive state information could be exposed, stored, or processed on external servers lacking statutory information safety controls.
Scope of Access Restrictions
The restriction applies directly to personnel within the Government Offices of Sweden, the core executive administrative organization supporting the prime minister and government ministries.
The mandate extends across a growing network of public institutions:
- Government Offices of Sweden: Web access to ChatGPT and Claude on work devices is restricted.
- Swedish Tax Agency: Parallel restrictions on public AI platforms have been put in place.
- National Board of Institutional Care: Operating under enforced restrictions on non-approved AI platforms.
- Swedish Enforcement Authority: Currently reviewing internal technical controls to determine whether to impose identical access blocks.
- Swedish Transport Administration: Previously conducted an internal review that flagged operational risks linked to AI-generated responses and the handling of confidential data.
Compliance and Approved Alternatives
The administrative mandate does not impose an outright ban on artificial intelligence tools across government functions, but shifts operational rely-on toward enterprise infrastructure with explicit data safeguards.
Government staff are required to use Microsoft Copilot for official duties, which remains authorized after completing internal security evaluations and subject to heightened technical access controls. Additionally, select personnel will maintain monitored access to ChatGPT and Claude through a custom internal AI assistant platform operated directly under government security supervision.
Mikael Wettercrantz, Director of Digitalisation at the Government Offices of Sweden, warned that deploying external commercial AI platforms without binding processing and protection agreements creates direct “consequences for our information security.”
