Why trust Greenbot
Elon Musk’s Department of Government Efficiency (DOGE) has deployed an artificial intelligence (AI) chatbot to take over tasks once performed by fired federal employees.
The chatbot, named GSAi, now serves about 1,500 workers at the General Services Administration (GSA).
GSAi’s rollout comes during massive job cuts across federal agencies. The GSA alone has lost hundreds of employees, including 90 percent of its tech team that once developed government software.
The AI tool handles basic office work: writing emails, creating talking points, summarizing documents, and coding. But current employees aren’t impressed.
“The chatbot is about as good as an intern,” one GSA worker told WIRED magazine. It produces “generic and guessable answers,” raising doubts about whether it can match human expertise.
The system faces strict limits on handling sensitive information. Workers cannot share nonpublic or “controlled unclassified information” with GSAi – a major problem since this represents much of the agency’s actual work.
Users can choose between three AI models: Claude Haiku 3.5 (the default), Claude Sonnet 3.5 v2, or Meta Llama 3.2.
GSA manages government buildings, technology infrastructure, and contracts nationwide. Under President Trump, the agency faces budget cuts up to 50 percent, forcing widespread staff reductions.
The plan to develop the government AI bot was first reported last month. Thomas Shedd, who heads GSA’s Technology Transformation Services, said the project was not new and has already been underway long “before we started.” Shedd also admits GSAi differs from previous government AI projects.
“The thing that’s different is potentially building that whole system in-house and building it very quickly,” he explained.
Early versions of similar government chatbots weren’t ready for deployment. Staff described previous prototypes as “janky,” yet DOGE pushed ahead with the rollout anyway.
Labor advocates warn that AI cannot duplicate the specialized knowledge of experienced civil servants. As automation expands across government, questions remain whether technology can deliver efficiency without sacrificing quality and institutional expertise.
