Anthropic Launches Claude for U.S. Civil Agencies

Two recent court decisions contradict each other regarding measures targeting Anthropic, while a meeting between Dario Amodei and Donald Trump has taken place. In this context, the publisher is deploying Claude for U.S. civil agencies, with a FedRAMP High environment and usage-based billing.

Opposing Court Decisions and Presidential Directive

At the end of September, a Washington appeals court upheld one of the designations against Anthropic. In late August, a judge in San Francisco had overturned the other designation, deeming it an illegal retaliation. Meanwhile, Donald Trump has ordered all agencies to abandon Claude. Anthropic's CEO, Dario Amodei, dined with Donald Trump on Sunday, which could indicate a de-escalation of tensions, although the situation remains uncertain.

Offering to Agencies and Control Guarantees

Since July, the government version of the Claude platform has been available in open beta and operates within a FedRAMP High-compliant infrastructure, which is the highest cloud security standard in the United States. Agencies have the same options as commercial clients and pay their bills based on usage, within predefined limits. Officials have the ability to allocate budgets and configure access to models by service, benefit from audit logs, and a dual-approval system for sensitive operations. Communications remain stored locally on the agencies' equipment.

Civil Scope and Defense Background

Anthropic is now providing its government version of Claude to federal agencies and U.S. state administrations, particularly civil entities. In March, the Pentagon banned the use of Claude after designating Anthropic as a threat to the supply chain, following its refusal to lift restrictions regarding mass surveillance and autonomous weapons. Despite this ban, the military used Claude during the war in Iran before discontinuing its use.