Sweden Blocks ChatGPT and Claude for Government Work Over Security Concerns

The Swedish government has blocked employees from using OpenAI’s ChatGPT and Anthropic’s Claude for official work due to data security fears, according to the publication. The two chat tools join DeepSeek on the nation's list of banned, non-approved artificial intelligence platforms.

Mikael Wettercrantz, the Chief Information Officer and Director of Digitization, explained that using these unregulated tools for government tasks creates a security risk called "Shadow IT". This happens because chat services used without an official enterprise contract can feed sensitive typed information right back to the developer, where it gets saved to train future models.

A review of AI tools used by the Swedish Transport Administration earlier this year uncovered problematic uses of these platforms. Without proper enterprise agreements, organizations also struggle to audit how AI tools are used. This creates data protection problems when protected information leaves Europe, such as when it is processed on a data center server located in the United States.

Sweden previously restricted DeepSeek over privacy violation concerns related to European privacy rules. Similar concerns led the United States to ban DeepSeek products because Chinese data laws could allow authorities access to company servers. The U.S. Department of Homeland Security also warned that the DeepSeek platform could contain malware and other security vulnerabilities.

Despite the Swedish restrictions, government workers can still use Microsoft Copilot for official duties, provided they follow much tighter security controls put in place after an internal review. A small group of employees will also maintain access to ChatGPT and Claude through a special platform built for AI assistants.

Summary

The Swedish government's decision highlights growing international concerns over how public agencies handle sensitive data when using popular artificial intelligence tools. By restricting unregulated platforms and requiring strict enterprise contracts, governments are trying to prevent confidential information from being stored and used by private AI developers.