valyzer
Politics · Security measure

ES Sweden's Government Offices block staff from the web versions of ChatGPT and Claude on work computers over data security concerns and direct them to Microsoft Copilot

The Government Offices of Sweden, which support the prime minister and government ministries, have blocked staff from accessing the web versions of ChatGPT and Claude on their work computers, with the restrictions taking effect on Monday 26 October, Gulf News reported on 27 October, citing Swedish public broadcaster Sveriges Radio. The stated concern is that sensitive information could be exposed through AI tools operating without approved data protection agreements. Mikael Wettercrantz, Director of Digitalisation at the Government Offices, warned that using AI platforms without formal agreements could have "consequences for our information security". Staff are directed to Microsoft's Copilot, which has undergone a security review and been approved for official use. The restrictions are not a complete ban: some employees keep access to ChatGPT and Claude through a specially developed AI assistant platform operated under government controls. Similar restrictions have been introduced at the Swedish Tax Agency and the National Board of Institutional Care, while the Swedish Enforcement Authority is reviewing whether to impose its own controls. The Government Offices had previously restricted access to DeepSeek.

Technical recordView JSON in the API (opens in a new tab)ID 85d5f4d9-15e3-4707-a73f-fc09df47494b

Place: Stockholm

View the country context in World →

Occurred
26 Oct 2026, 00:00 UTC
Last updated by Valyzer
Severity
Low
Confidence
50%
Verification
Automated verification (no editorial review)
Quality
No warnings
Evidence pieces
1

Confidence (50 %) estimates the strength of the reading; it does not substitute for verification.

Related entities

Facts

No claims beyond the headline have been published.

Relationships and impacts

Observed connections are separate from inferred impact mechanisms.

Observed relationships

No evidenced relationships have been published for this event.

Inferred impacts

Potential scenarios, not confirmed facts.

Documented participation; impact undetermined

These records identify participants in the act. They do not establish a positive or negative consequence on their own.

Sweden

Evidence for participation

Inference · Undetermined

Mikael WettercrantzNamed in the documented event; it does not establish a role, a favourable or adverse impact or its consequencesUndetermined

Horizon immediate · Confidence 50%

Evidence for the inferred impact

Accessible causal chain
EntityMechanismResultHorizonConfidence
Mikael WettercrantzNamed in the documented event; it does not establish a role, a favourable or adverse impact or its consequencesUndeterminedimmediate50%

Sources consulted

Deduplicated evidence pieces behind this event. Sources are repeated only in the blocks that cite them.

Evolution and evidence

Browse what changed and what Valyzer knew at the selected cut. Comparisons use explicit versions, not page order.

Valid at
Known until
Reset
  1. Publication

    Recorded

    Public reason: Public reason unavailable

    Validity
    — current