Sweden blocks OpenAI and Anthropic AI tools in government offices

Swedish government bans external AI tools

Sweden's government offices have blocked access to the popular AI assistants ChatGPT and Claude. The move restricts the use of external services from OpenAI and Anthropic within official government channels. Instead, the administration is retaining access to Microsoft Copilot, an AI tool integrated into Microsoft Office products.

The decision was reported by Decrypt on October 24, 2026. This shift highlights a preference for controlled, enterprise-grade AI solutions over open-ended consumer-facing chatbots in the public sector.

Key changes in AI policy

• ChatGPT (made by OpenAI) is no longer accessible in Swedish government offices.
• Claude (made by Anthropic) is also blocked from use in these offices.
• Microsoft Copilot remains an approved tool for government staff.

What the reporting says

According to Decrypt, the blocking of these specific AI assistants is part of the Swedish government's broader management of digital tools. While the source does not detail the specific security or data privacy reasons behind banning ChatGPT and Claude, the retention of Microsoft Copilot suggests a strategy to use AI within established software ecosystems. Microsoft Copilot is an AI assistant built into Microsoft's productivity applications, allowing users to generate text, summarize data, and automate tasks without leaving their existing work environment.

Why it matters

This move is significant because it shows how governments are curating their AI tools for security and control. By restricting external chatbots and keeping a bundled solution, the Swedish government aims to manage data flow and ensure that AI usage aligns with public sector standards. It reflects a growing trend among institutions to prefer integrated, managed AI platforms over standalone third-party applications, especially in sensitive environments like government offices.

Sources

• Decrypt