Skip to content
AI EmployeeSuper-AgentsAgent-to-AgentBenchmarksPricingBlog

Sweden Blocks ChatGPT and Claude on Government PCs

At a glanceQuick answers
Did Sweden ban ChatGPT and Claude?
No. Its Government Offices block the two websites on staff work computers from October 26; some staff keep the models through the Intric platform.
Why?
Using AI services without an agreement can affect information security, the offices’ digitalisation director wrote on the intranet.
What can staff use instead?
Microsoft Copilot, approved for all staff after a security review, and for some staff the ChatGPT and Claude models inside Intric.
Data illustration on a near-white ground: a government building beside a laptop whose two browser tabs are closed off by a gate, with an open side path leading to a second door, under the large date Oct 26
Fig 0From Monday, the public websites close and the contracted route stays open. Made by CellCog's image agent, running GPT Image 2.5.

Sweden’s Government Offices will block the ChatGPT and Claude websites on staff computers from Monday, October 26, 2026. Sveriges Radio’s Ekot reported the decision at 10:20 UTC on October 23, and the offices confirmed it to SVT by email. Staff are pointed to Microsoft Copilot instead, and some keep both models through a contracted platform.

It is not a ban on the models. It closes the public websites on work machines, where staff were using AI services with no agreement behind them. Ekot found the same move at two more agencies, and a third is weighing it.

On this page · 8 sectionsOpen
  1. What changes on Monday
  2. Why: AI services with no agreement
  3. Other Swedish agencies
  4. Intric: the route that stays open
  5. What it means for teams using AI at work
  6. Where CellCog stands
  7. What we are watching
  8. Sources
Key points5 · 5 min full read
  1. A browser window with a closed gate across it: the block.
    Sweden’s Government Offices block the ChatGPT and Claude websites on staff work computers from Monday, October 26, 2026, Ekot reported on October 23.
  2. An arrow turning to an approved door: the redirect.
    Staff are directed to Microsoft Copilot, which the offices approved and made more secure after a review.
  3. A second, open door beside the closed one: the route that stays.
    Some staff keep the ChatGPT and Claude models through Intric, a Swedish AI platform the offices procured in July 2026.
  4. A padlock on a document: information security.
    The reason given: using AI services without an agreement can harm information security, wrote digitalisation director Mikael Wettercrantz.
  5. Three small buildings, two with a gate: other agencies.
    Ekot found blocks already at the Tax Agency and SiS, and the Enforcement Authority is preparing a position.

§ 01What changes on Monday

The Government Offices (Regeringskansliet) told SVT that the browser interfaces for ChatGPT and Claude will no longer be reachable from its service computers from October 26. Every employee has Microsoft’s AI tool, and inside Intric, employees still have access to the ChatGPT and Claude language models.

Animation: a gate marked Oct 26 drops across the paths from a work computer to the ChatGPT and Claude websites, while paths to Copilot and to Intric, which reaches the ChatGPT and Claude models, stay open
Fig 1From Oct 26 the two websites close; Copilot and Intric stay open
Route Status from October 26 Source
ChatGPT website on a work computer Blocked Offices’ email to SVT
Claude website on a work computer Blocked Offices’ email to SVT
Microsoft Copilot Approved for all staff; security raised after a review Ekot, SVT
ChatGPT and Claude models inside Intric Still available to some staff Ekot, SVT
DeepSeek Already restricted before this change Ekot
Table 1What changes at Sweden’s Government Offices on October 26, 2026 (SVT, Ekot)

§ 02Why: AI services with no agreement

Digitalisation director Mikael Wettercrantz declined an interview with Ekot. On the offices’ intranet he wrote that using AI services without an agreement can have “konsekvenser för vår informationssäkerhet”, consequences for information security.

Ekot points to a concrete case. Earlier this year an internal audit at Trafikverket, the transport agency, found widespread and risky use of about ten AI services, with problems both in how sensitive the information typed in was and in how the answers were handled.

§ 03Other Swedish agencies

Agency What Ekot reported
Government Offices (Regeringskansliet) ChatGPT and Claude websites blocked from Oct 26; DeepSeek restricted earlier
Swedish Tax Agency (Skatteverket) Block already in place; services and start date not stated
National Board of Institutional Care (SiS) Block already in place; services and start date not stated
Swedish Enforcement Authority (Kronofogden) Preparing a position
Swedish Transport Administration (Trafikverket) Internal audit found risky use of about ten AI services
Table 2Swedish agencies and public AI services, as reported by Ekot on October 23, 2026

Ekot did not say when the Tax Agency and SiS blocks began or which services they cover, so treat those two rows as the least specific.

Animation: five Swedish agency buildings; bars drop across the doors of the Government Offices, the Tax Agency and SiS, the Enforcement Authority is preparing a position, and the Transport Administration shows an audit of about ten AI services
Fig 2Two more agencies already block; one is preparing a position

§ 04Intric: the route that stays open

Intric is a Stockholm AI platform company. On July 8, 2026 it announced it had won the Government Offices’ procurement for a platform where staff build and run AI assistants on the offices’ own data. In Intric’s words, “The platform runs on Swedish infrastructure without dependence on foreign cloud services.” The agreement covers both a cloud service and an on-premises installation.

That is the pattern: keep the models, change the door. The public websites close; a contracted platform with its own governance stays open. Neither the offices nor Intric has said publicly which deployment the offices run, or where requests to the ChatGPT and Claude models are processed.

§ 05What it means for teams using AI at work

Sweden’s move is a template other organizations will copy. The question it asks is simple: which AI tools are approved, and under which agreement? Teams that can answer it keep their tools. Teams running on personal accounts with nothing signed are the ones that get a block on a Monday morning.

§ 06Where CellCog stands

Every CellCog agent tier runs on Claude Opus 5.5, reached through Anthropic’s API, not through Claude’s website. Sweden’s block targets the websites, not the models, and its own approved route reaches the same models through a contracted platform.

§ 07What we are watching

  • Whether the Tax Agency and SiS say what they blocked, and since when.
  • The Enforcement Authority’s decision.
  • Whether other Nordic governments close the public websites the same way.

§ 08Sources

Frequently asked5 questions

Q1When does Sweden's block take effect?

Monday, October 26, 2026, according to the Government Offices’ email to SVT.

Q2Is it a ban on the ChatGPT and Claude models?

No. It blocks the public websites from service computers. The offices told SVT that staff still reach both models inside Intric.

Q3Which other Swedish agencies blocked AI services?

Ekot reported blocks at the Swedish Tax Agency and the National Board of Institutional Care, and said the Enforcement Authority is preparing a position.

Q4What is Intric?

A Stockholm AI platform company that won the Government Offices’ procurement in July 2026. It says its platform runs on Swedish infrastructure, as a cloud service or on premises.

Q5Does CellCog run on Claude?

Yes. Every CellCog agent tier runs on Claude Opus 5.5, reached through Anthropic’s API rather than Claude’s website.

Published 25 October 2026 All Trust, permissions & security →