In the absence of a data protection contract, the Swedish Government Offices cut off access to the general public versions of ChatGPT and Claude and direct their agents to Microsoft Copilot.
Since Monday, October 26, agents of the Government Offices, the administration that supports the Prime Minister and the Swedish government, can no longer open ChatGPT or Claude from their work computers. The director of digital affairs, Mikael Wettercrantz, justifies the measure by information security: these consumer services, used without a contractual agreement, send the data entered to foreign servers where American law allows the authorities to require its communication. The administration says it has observed problematic uses involving sensitive information.
Microsoft Copilot becomes the only approved assistant, after a security review and a tightening of its parameters; controlled internal AI services also remain accessible. Sweden follows several of its agencies, including the tax administration and the collection authority, which had already restricted these tools, while an audit of the transport administration had identified a dozen AI services used in a "broad and risky" way. The approach remains less radical than that of the European Parliament, which had disabled all AI functions, including Copilot, in February. Ironically, Copilot has been integrating Anthropic models itself since the beginning of 2026, even though they are disabled by default in European regions.
Read the full article on the original source
Go to the source →