[DigitalToday reporter Choi Jae-won] Anthropic has officially launched Claude for Government, an AI service targeting U.S. government agencies.

On Oct. 1 (local time), IT outlet TechRadar reported that Claude for Government has now shifted to a full service after being introduced as a public beta in July. It is a FedRAMP High-approved SaaS for U.S. federal agencies and regulated environments, designed to allow Claude to be used for sensitive but unclassified work.

Government agency staff can use functions similar to the commercial Claude. Anthropic plans to apply new features to the government version on a cadence similar to its commercial service. Administrators can restrict available models, check usage by user and control external connections.

The billing model has also changed. Instead of a per-seat fee, agencies allocate a set amount in advance and use the service within that limit. Administrators can set spending caps and model access levels by department or user group and track usage by user and model. It also provides depletion warnings before balances run low. The move is intended to prevent sharp increases in AI spending at public institutions and make budget execution easier to manage.

Anthropic also expanded Claude for Microsoft 365 to early access. Users can use Claude directly in Word, Excel and PowerPoint, while Outlook support is in beta. Claude had already been integrated with Microsoft 365 to analyze documents, emails and collaboration data.

The launch shows that competition among generative AI companies in the U.S. public sector is expanding beyond model performance to include security certifications, budget controls and the ability to integrate with existing work tools.