Sweden’s Government Offices Block ChatGPT and Claude, Steer Staff to Copilot

Sweden’s Government Offices, the administrative body that supports the prime minister and cabinet, will block ChatGPT and Claude on staff computers starting Monday, Oct. 26, 2026, and direct civil servants to Microsoft Copilot instead. The decision to make Sweden block ChatGPT and Claude was first reported by public broadcaster Sveriges Radio’s Ekot and confirmed by the Government Offices in an email to SVT.

Image: Rosenbad, the seat of the Swedish government, on the Norrström waterfront in Stockholm. Photo: trolvag, via Wikimedia Commons (CC BY-SA 3.0).

What the Government Offices is blocking, and when

In its email to SVT, the Government Offices (Regeringskansliet in Swedish) said the browser interfaces for ChatGPT and Claude “will no longer be reachable from the Government Offices’ work computers as of Oct. 26.” All employees will still have access to Microsoft’s AI tool, Copilot, which the Government Offices approved after reviewing the service and raising its security settings, SVT reported.

The ban is not absolute. The Government Offices told SVT that staff can still use the ChatGPT and Claude language models through Intric, an internal AI platform. Decrypt reported that the press office has not said how many people have that access or what limits apply to it.

Information security behind Sweden’s ChatGPT and Claude block

The reasoning came from Mikael Wettercrantz, the Government Offices’ director of digitalization, in a post on the staff intranet cited by Sveriges Radio and Decrypt. He wrote that using AI services without a formal contract can have “consequences for our information security.”

“We have seen several examples of problematic use of open AI services,” Wettercrantz said, according to Decrypt’s translation. “The examples concern both the sensitivity of the information that has been entered and how the AI-generated responses have been handled.”

The core issue is contracts. Consumer AI tools without a formal agreement send what users type to the provider’s servers, and a vetted enterprise agreement at least sets out what the provider may and may not do with government employees’ inputs. Decrypt also noted that U.S. law allows authorities to compel American companies to hand over data, a long-running concern for European governments that rely on U.S. cloud services.

Other Swedish agencies already moved

The Government Offices is not the first Swedish public body to act. SVT reported that the Swedish Tax Agency (Skatteverket) and the National Board of Institutional Care (SiS) had already introduced similar blocks, and that the Swedish Enforcement Authority (Kronofogden), which handles debt collection, is preparing its own decision. Decrypt added that an internal audit at the Swedish Transport Administration found roughly 10 AI services in use there in a “widespread and risky way.” Computer Sweden reported that the Government Offices had also previously restricted use of the Chinese AI service DeepSeek.

Sweden is drawing a narrower line than Brussels. Decrypt reported that the European Parliament disabled AI features on all its work devices in February, including ChatGPT, Claude and Copilot, after concluding it could not guarantee the security of the data those tools processed. European unease about AI and data protection dates back at least to Italy’s temporary ChatGPT ban in 2023.

A wrinkle: Claude inside Copilot

The split between the “banned” and “approved” tools is less clean than it looks. Microsoft began offering Anthropic’s Claude models inside Copilot in early 2026, and in the European Union those models are switched off by default and must be enabled by administrators, Decrypt reported. The Swedish Government Offices has not said whether Claude is turned on in its Copilot setup.

The Claude question points to the real dividing line: not which AI model a civil servant uses, but whether the government has reviewed and contracted the service it runs through. It is the same logic behind a recent move in Washington, where the Department of Homeland Security ordered federal agencies to remove DeepSeek AI from their systems, though that order targeted a Chinese company over national security concerns rather than U.S. providers over contract terms.

What it means for OpenAI and Anthropic

For the companies behind ChatGPT and Claude, government customers are a growing prize, and Microsoft’s existing enterprise contracts give it an edge with public agencies that already run its software. OpenAI has been widening its cloud partnerships, as we reported when OpenAI models came to Google Cloud as Microsoft’s exclusivity faded, and Anthropic is preparing for a public listing, with its S-1 draft showing $16.3 billion in revenue. Neither company had publicly commented on the Swedish decision in the reports reviewed by The Boston Report.

The Swedish block takes effect Monday. Whether other European governments follow with similar Copilot-only rules, or with broader bans like the European Parliament’s, will be worth watching as agencies across the continent write their first formal AI policies.

Sources

• SVT Nyheter: AI-tjänster spärras på Regeringskansliet (in Swedish)
• Decrypt: Sweden’s Government Offices block ChatGPT and Claude, keep Microsoft Copilot
• Computer Sweden: Regeringskansliet spärrar Chat GPT och Claude (in Swedish)
• Wikimedia Commons: Rosenbad, Stockholm, photo by trolvag (CC BY-SA 3.0)