The Swedish government has restricted access to ChatGPT and Claude on official work computers, citing concerns about information security and the risks associated with using artificial intelligence tools without formal data-protection agreements.

The restriction took effect on Monday, October 26, 2026, and applies to browser-based versions of OpenAI’s ChatGPT and Anthropic’s Claude used by staff at Sweden’s Government Offices, which supports the prime minister and government ministries.

The move does not amount to a nationwide ban on the two AI platforms. Instead, it establishes tighter rules on which tools government employees can use when carrying out official duties.

Why Sweden Is Restricting ChatGPT and Claude

The main concern is how sensitive government information could be handled when employees enter it into AI services that have not been formally approved for official use.

Government officials have warned that using external AI platforms without appropriate agreements could expose public-sector information to security and data-management risks.

Mikael Wettercrantz, the Government Offices’ director of digitalisation, has said that the use of AI services without agreements can have consequences for information security.

The concerns follow an internal review of AI usage at the Swedish Transport Administration earlier in 2026. The review identified problematic practices involving both the sensitivity of information submitted to AI tools and the handling of AI-generated responses.

The issue highlights a challenge facing public institutions as AI becomes more widely used in everyday office work: employees may find these tools useful, but organisations must ensure that their use complies with security requirements and data-protection policies.

Microsoft Copilot Remains Available

Rather than abandoning AI altogether, Sweden’s Government Offices is directing employees towards approved alternatives, including Microsoft Copilot.

The government has reviewed Copilot and strengthened its security settings, allowing staff to continue using an authorised AI assistant under more controlled conditions.

The distinction is not simply about which chatbot offers better features. It is also about the agreements, safeguards and administrative controls that determine how an organisation can manage the information employees share with an AI service.

Some government employees will also retain access to ChatGPT and Claude through a dedicated AI-assistant platform developed for official use. This means the restriction targets direct access through ordinary web browsers rather than eliminating every possible route to the underlying AI models.

Other Swedish Agencies Are Taking Similar Measures

The restrictions extend beyond the central government offices.

Sweden’s Tax Agency and the National Board of Institutional Care have also introduced restrictions on certain AI services. The Swedish Enforcement Authority has been reviewing the issue as it considers its own approach.

These developments suggest that public institutions are increasingly examining how staff use AI tools and whether existing arrangements provide adequate protection for official information.

The challenge is to establish rules that allow employees to benefit from AI-assisted work without creating unnecessary exposure of confidential documents, personal information or sensitive government material.

What the Decision Means for the Future of AI

Sweden’s decision illustrates an important distinction in the growing debate over artificial intelligence: restricting an AI service for government work does not necessarily mean rejecting the technology itself.

Public institutions can continue exploring AI for drafting, research, administrative support and other tasks while limiting the use of services that have not met their security and contractual requirements.

For technology companies, the development also highlights the importance of providing organisations with clear data-handling arrangements, appropriate security controls and systems that meet public-sector requirements.

As governments introduce more detailed AI policies, access to these tools may increasingly depend on how they are deployed, the protections attached to their use and the type of information employees are permitted to process.

For Sweden, the immediate priority is to ensure that civil servants use AI within an approved framework while protecting the information entrusted to government institutions.

GhanaNews24.com will continue to follow developments in artificial intelligence, digital security and technology policy around the world.