Sweden's Government Offices Restricts AI Tools Over Security Concerns, Approves Microsoft Copilot

Sweden’s Government Offices will block the use of ChatGPT and Claude on staff computers starting Monday, designating Microsoft Copilot as the sole approved artificial intelligence tool due to information security concerns, according to Mikael Wettercrantz, the Government Offices’ director of digitalisation.

“Using AI services without a formal contract can have consequences for our information security,” Wettercrantz stated on the staff intranet. “We have seen several examples of problematic use of open AI services. The examples concern both the sensitivity of the information that has been entered and how the AI-generated responses have been handled.” He explained that consumer AI tools without formal agreements may send user queries to external company servers, where U.S. law could compel data disclosure. A formal security agreement, he noted, defines what a provider can and cannot do with government employees’ inputs, distinguishing vetted enterprise contracts from open consumer tools.

Microsoft Copilot was approved after the Government Offices conducted a security review and adjusted its settings, a step not taken for ChatGPT or Claude. Despite the restrictions, some employees may still access the blocked tools via an internal AI platform developed by the Government Offices. The press office has not specified how many employees this covers or what restrictions apply to this internal platform.

These measures align with a broader trend among Swedish government entities and international bodies addressing AI security. Sweden’s Tax Agency and the National Board of Institutional Care have already implemented similar bans on consumer AI tools. The Enforcement Authority is also preparing its own decision, and an internal audit at the Transport Administration revealed approximately 10 AI services being used in a “widespread and risky way.”

Internationally, the European Parliament took a more extensive step in February by disabling AI features across all its work devices, including ChatGPT, Claude, and Copilot, citing an inability to may provide the security of data processed by these tools. European government concerns about AI data security have been escalating since Italy’s ban on ChatGPT in 2023. Sweden’s current approach appears to draw a narrower line than Brussels by sanctioning one U.S. tech company’s product while restricting others.

However, the distinction between the “banned” AI and the “approved” one may not be entirely clear-cut. Microsoft incorporated Claude models into Copilot in early 2026. Within the EU, Anthropic’s models (which power Claude) default to being off and require administrators to enable them deliberately. Sweden’s Government Offices has not reported whether its specific Copilot instance has Claude switched on, introducing a layer of uncertainty regarding the absolute separation of the tools.

The decision by Sweden’s Government Offices underscores the ongoing challenge for governmental bodies to balance the adoption of advanced AI technologies with the imperative of safeguarding sensitive information. The focus remains on establishing clear security protocols and contractual agreements to manage the risks associated with AI tool usage.

Why This Matters

The materials describe a narrow update: Sweden’s Government Offices has blocked the use of ChatGPT and Claude on staff computers, designating Microsoft Copilot as the sole approved AI tool. How many employees can access blocked AI tools through the internal AI platform.

Broader Context

Source materials place the factual news in this context: Sweden’s Government Offices—the administrative body supporting the prime minister and cabinet—will block ChatGPT and Claude on staff computers from Monday, directing civil servants to Microsoft’s Copilot as the only approved AI tool, according to Sweden’s national, publicly funded radio broadcasterSveriges Radio.