Effective October 26, 2026, Sweden's Government Offices blocked staff from accessing browser-based versions of OpenAI's ChatGPT and Anthropic's Claude on work devices due to data security risks. Digitalisation Director Mikael Wettercrantz cited concerns over processing sensitive government data without formal data-protection agreements. Staff are directed to use Microsoft Copilot, while limited access to ChatGPT and Claude remains through Intric, an internal platform hosted on Swedish servers.
Ban on ChatGPT and Claude on government computers
- ▪Sweden's Government Offices blocked staff from accessing web browser versions of OpenAI's ChatGPT and Anthropic's Claude on work computers effective October 26, 2026.
- ▪Sveriges Radio first reported the plan by Sweden's Government Offices to block ChatGPT and Claude on October 23, 2026.
Security concerns over unapproved AI platforms
- ▪Mikael Wettercrantz, director of digitalisation at Sweden's Government Offices, warned that using unapproved shadow IT platforms without security agreements threatens information security.
- ▪Sweden's Government Offices instituted the AI restrictions due to information security concerns over processing sensitive government data without formal data protection agreements.
- ▪An internal review at the Swedish Transport Administration earlier in 2026 found problematic and risky uses of approximately 10 unapproved AI services by government workers.
- ▪Swedish officials expressed concern that entering sensitive data into unapproved AI platforms could cause information to be stored or used to train public models.
Transition to approved Microsoft Copilot service
- ▪Sweden's Government Offices directed staff to use Microsoft Copilot after it passed an internal security screening and underwent strengthened security settings.
- ▪Microsoft Copilot operates under a security agreement between Microsoft and the Swedish government governing data control and settings.
Secure access via Swedish Intric platform
- ▪Intric, a Stockholm-based firm that won a contract with Sweden's Government Offices in July, operates its AI assistant platform on domestic Swedish servers.
- ▪A select number of staff at Sweden's Government Offices can still access OpenAI and Anthropic models through Intric, an internally managed Swedish AI assistant platform.
AI restrictions across other Swedish agencies
- ▪The Swedish Tax Agency and the National Board of Institutional Care have also implemented blocks on unapproved public AI services on work computers.
- ▪The Swedish Enforcement Authority is reviewing whether to restrict public AI services on its employee work computers.
Story comments
Loading comments…