AI Is Here, It’s Time Governments Stop Resisting and Start Leading Responsibly
the GSA requires the agencies to report usage and alert the GSA to any security or ethics problems, and it keeps the program open to new vendors
The effort is a follow-up to the White House June "Blueprint for Responsible AI," which challenged agencies to pilot machine-learning answers in healthcare diagnosis, infrastructure design and public-sector analytics, provided that each project includes robust impact assessments. Photo/Courtesy
By Mercy Chelangat
The U.S. General Services Administration on Tuesday, August 5, 2025 approved OpenAI’s ChatGPT, Google’s Gemini and Anthropic’s Claude as pre-qualified AI vendors for federal agencies, streamlining procurement under unified contracts that mandate data-privacy, security and ethical-use standards from the outset.
By pre-clearing the three locations, the GSA cuts months of horse-trading on individual contracts, so any agency; from Veterans Affairs to the Environmental Protection Agency; can gain access to the latest AI tools within hours. The change promises quicker deployment of applications like virtual customer-service agents, predictive maintenance on government buildings and automated fraud detection, all to the same rigorous protocols to remove bias and protect against sensitive information.
The effort is a follow-up to the White House June “Blueprint for Responsible AI,” which challenged agencies to pilot machine-learning answers in healthcare diagnosis, infrastructure design and public-sector analytics, provided that each project includes robust impact assessments. “Pre-qualifying these providers puts innovation at the speed of need,” said GSA Administrator Robin Carnahan, noting that the list of vendors will be re-reviewed quarterly and expanded with other companies with comparable requirements.
Others caution that centering federal AI spending on three big companies could limit competition. In a precautionary measure, the GSA requires the agencies to report usage and alert the GSA to any security or ethics problems, and it keeps the program open to new vendors who fulfill the stipulated requirements.
Elsewhere, other governments are taking notice too. Brussels’ European Commission is racing to finish its AI procurement policy within the forthcoming AI Act, and Canada’s Treasury Board Secretariat is piloting a similar pre-approved vendor list for healthcare and transportation usage.
As nations battle for AI supremacy, they have a choice: cling to fear-mongering prohibition or embark on thoughtful adoption. The GSA’s action on August 5 demonstrates the way policy can facilitate forward momentum and ensure public trust. It is time for nations to come forward and regard AI not as a force to be feared and kept at arm’s length, but as a powerful tool to be directed with vision and caution.
About Author
