The United States Department of State is reportedly the latest government agency to move away from Anthropic’s Claude AI model. A report claims that the US government has replaced the AI model powering its internal chatbot, StateChat. According to an internal document seen by NextGov/FCW, the agency has moved away from Anthropic’s Claude Sonnet 4.5 model to OpenAI’s GPT-4.1. The change follows US President Donald Trump’s directive instructing federal agencies to remove Anthropic tools from their systems.
The document confirms that Anthropic’s software is no longer powering StateChat. It also notes that the transition to a different large language model provider has affected the chatbot’s data, resetting it to information available as of May 2024.
Earlier, the StateChat system running on Claude had access to more recent training data from June 2025, according to a source familiar with the matter who spoke to Nextgov/FCW on condition of anonymity.
The internal document also states that the US State Department employees using custom GPT systems built on Claude were instructed to migrate to another government-approved model which was not developed by Anthropic, earlier this month.
What US State Department said about Anthropic models in its tools
In a statement to Nextgov/FCW, a spokesperson for the United States Department of State confirmed the change said,
Reuters previously reported that several federal agencies, including the State Department, had received directives requiring them to move from Claude to ChatGPT.
Claude had earlier been introduced for federal agency use through the General Services Administration OneGov programme, which negotiated software pricing arrangements for government departments, often for limited periods.
Following the government-wide restriction on Anthropic’s technology, the company filed two lawsuits this week. The case filed in the US Court of Appeals for the DC Circuit challenges the government’s designation of Anthropic as a under provisions of the Federal Acquisition Supply Chain Security Act of 2018.
The second case, filed in the US District Court for the Northern District of California, names several government agencies and officials as defendants and seeks injunctive relief against retaliation against Anthropic.
