From Monday, 26 October 2026, staff at Sweden’s Government Offices can no longer open the ChatGPT or Claude websites from their work computers. The change was first reported by Sveriges Radio’s news desk Ekot on 23 October, and the Government Offices confirmed it to broadcaster SVT: the browser interfaces for ChatGPT and Claude will no longer be reachable from staff work computers from 26 October.
What happened
The Government Offices (Regeringskansliet) is the administrative body that supports Sweden’s prime minister and cabinet. Its new rule covers the public chatbot websites on managed work computers — not the AI models themselves, and not the whole country.
Three points keep the story accurate. First, Microsoft Copilot stays approved for all staff, after the offices reviewed it and raised its security settings. Second, some staff still reach the ChatGPT and Claude models through Intric, an internal AI assistant platform, so the models have not vanished from government work entirely. Third, this is not the first such restriction in Sweden: the Government Offices had already restricted the Chinese AI service DeepSeek, and Ekot found similar blocks already in place at the Swedish Tax Agency and the National Board of Institutional Care, while the Enforcement Authority is still weighing the question.
The stated reason is information security, not a single incident. Digitalisation director Mikael Wettercrantz wrote on the offices’ intranet that using AI services without an agreement can have consequences for information security. No public statement in the reporting says a leak prompted the move.
Why it matters for small businesses
Your team probably does not work for a government, but the logic behind this decision applies to almost every small business. A public chatbot is convenient, and that convenience is exactly the risk: anything an employee pastes into it — a client contract, an internal price list, an unreleased product description — leaves your systems and is handled under the provider’s standard terms.
Sweden’s answer was not to ban AI. It was to draw a line between approved tools and unapproved ones, and to make the approved route easy enough that staff actually use it. That is a policy any small team can copy, and it costs nothing to write down.
The timing also matters. Clients, partners and regulators are all asking harder questions about where business data goes. A one-page AI-use policy — which tools are approved, what must never be pasted into a public chatbot, and who to ask — is now a basic credibility signal, not a bureaucratic extra.
What to do next
- Pick one approved AI tool for everyday work, and write it down. If your team already uses Microsoft 365, an approved Copilot setup is the path of least resistance; if not, choose one business plan with admin controls rather than letting everyone use personal accounts.
- Ban the risky inputs, not the tool. Forbid customer personal data, passwords, financial details and unreleased business information in any public chatbot, approved or not.
- Offer a private alternative for sensitive documents. A self-hosted or internal chatbot keeps document Q&A inside your own systems — see our guides to a private AI chatbot for business documents (https://leanstackai.com/anythingllm-private-ai-chatbot-tutorial/) and to running LLMs privately on business PCs (https://leanstackai.com/text-generation-webui-tutorial/).
- Tell the team in plain language. A policy nobody has read changes nothing. Share the approved list, the banned inputs and the reason in one short message, and repeat it when someone new joins.
- Review quarterly. Approved tools change their terms, new tools appear, and today’s safe default may not be tomorrow’s. Put a 15-minute AI-policy check on the calendar every three months.
Sweden’s move is a reminder, not a reason to panic. The businesses that handle this well will not be the ones that ban AI outright — they will be the ones that decide, in advance, which tools are approved and what data stays out of them.
Frequently asked questions
Is Sweden banning ChatGPT and Claude for everyone?
No. The block covers the public ChatGPT and Claude websites on Government Offices work computers from 26 October 2026. It is a workplace rule for one government body, not a nationwide law, and it does not stop Swedish citizens using either product on personal devices.
Why is Microsoft Copilot still allowed?
The Government Offices reviewed Copilot and raised its security settings before approving it for all staff. The deciding factor reported by Ekot is a formal agreement: using AI services without a contract was flagged as an information-security risk.
Should my small business copy this policy?
Copy the principle, not the exact list. Approve one or two tools with business controls, ban sensitive inputs in public chatbots, and give staff a private alternative for confidential documents. That approach keeps the productivity of AI without handing your business data to a public service on standard consumer terms.
Sources
- Gobi Daily, ChatGPT and Claude Blocked on Sweden’s Government PCs From October 26: https://gobidaily.com/articles/chatgpt-and-claude-blocked-on-swedish-government-pcs/
- Notebookcheck, Sweden’s government blocks ChatGPT and Claude, keeps Copilot: https://www.notebookcheck.net/Sweden-s-government-blocks-ChatGPT-and-Claude-keeps-Copilot.1433522.0.html
- CoinCentral, Sweden Blocks ChatGPT and Claude for Government Staff, Keeps Microsoft Copilot: https://coincentral.com/sweden-blocks-chatgpt-and-claude-for-government-staff-keeps-microsoft-copilot/
- ExplainX, Sweden Blocks ChatGPT and Claude From Oct 26 (Ekot): https://explainx.ai/blog/sweden-government-offices-block-chatgpt-claude-work-computers-copilot-ekot-2026
Original reporting: Sveriges Radio Ekot, 23 October 2026, confirmed by the Government Offices to SVT.
