US Government Launches AI Sandbox for Federal Agencies, Signaling Short-Term Commitment

GSA launches AI sandbox, says it won't be around for long

GSA AI chief says market can do better but the feds needed a kickstart The Trump administration just launched a detailed, AI-pushing platform for federal agencies last week, but a government leader is already promising to kill it. USAi.gov, the site launched by federal government contractors last week, appears to be the final form of the hub that was publicly exposed on GitHub in June. That leaked site revealed the Trump team's plans for a resource that would simplify AI adoption by federal agencies. The USAi name was originally supposed to reside at ai.gov, but the latter site now serves the purpose of talking up Trump's AI action plan. Compare a cached copy of the old ai.gov we reported on in June to the live USAi.gov and the similarities are apparent, both in the text and graphics, portions of which are identical. USAi is now a full-fledged cloud-based, shared AI evaluation platform where government agencies can experiment with different models in a sandbox to find what works for their particular needs. Because resources on USAi are shared, the GSA said, agencies won't have to duplicate work done by someone else if it's useful for their needs. In addition to the evaluation suite, USAi also includes a dashboard and analytics platform that allows agencies to track their work on the platform to see what, statistically, performs best for them. Models from OpenAI, Google, and Anthropic were available at launch with plans to add more as they get government approval. "USAi provides a secure and collaborative environment for government employees to explore cutting-edge AI models that will make day-to-day workflows more efficient and help discover innovative solutions prior to making procurement decisions," Federal Acquisition Systems commissioner Josh Gruenbaum said last week. That could be a worthwhile investment for the federal government. As GSA CIO David Shive put it in the administration's statement, the public pooling of such early AI experimentation will save a lot of money and hopefully make the process more efficient for agencies. That doesn't mean USAi is going to be a permanent fixture. GSA Chief Data and AI Officer Zach Whitman told attendees at a Salesforce event in Washington, D.C. on Tuesday that the federal government wants to make sure it's not in this business for the long-term. The event was sponsored by FedScoop, which first reported Whitman's comments. The market will eventually address the need instead, Whitman opined. But the space needs a bit of a government kickstart to get agencies past any acquisiton or technical barriers. Whitman also opined that USAi's shared service design is an early if necessary limitation, as it won't allow agencies with niche AI needs to explore how they could use available tools. "There are a ton of really niche use cases that would require specific adversarial models or specific models that would not fit with the general-purpose platform," Whitman said in DC on Tuesday. "But again, the main focus is to allow for agencies to get up to speed quickly at a low to zero cost, and then allow the market to take over." With such an attitude from a GSA leader toward USAi's launch, one wonders how long it will last - as well as how much it cost to build. We reached out to the GSA for comment, but didn't hear back by press time. ®

Washington Opens AI Sandbox for Federal Agencies | PYMNTS.com

By completing this form, you agree to receive marketing communications from PYMNTS and to the sharing of your information with our sponsor, if applicable, in accordance with our Privacy Policy and Terms and Conditions. Through USAi.gov, federal employees can log in and try out AI tools in a secure space that complies with regulations. These tools include AI chatbots, coding assistants and AI that provides document summaries. The tools are in the cloud, not on employee computers. This shared sandbox lets government employees safely try out AI tools before deciding if they want to adopt it. By centralizing experimentation, GSA officials said the platform reduces duplication, strengthens security and speeds adoption of AI across government. "USAi means more than access -- it's about delivering a competitive advantage to the American people," GSA Deputy Administrator Stephen Ehikian said in the release. The move comes as the federal government seeks to scale AI responsibly while keeping pace with private sector advances. With USAi, agencies now have access to a centralized testing ground as Washington seeks to cement U.S. leadership in AI innovation. The agency said USAi supports the priorities in President Trump's America's AI Action Plan, a national strategy to strengthen U.S. leadership in AI through coordinated federal action, responsible innovation and modernized infrastructure. Released earlier this year, the plan calls for less regulation, wider AI adoption across agencies, workforce training and safeguards to ensure transparency and accountability. State and local governments are also working on adopting AI. According to a 2024 study by the National Association of State Chief Information Officers and McKinsey, 92% of states are developing or working on gen AI frameworks and policies. Eight out of 10 states are working with AI industry experts and researchers, while 53% are investing in training programs to help their staff prepare for gen AI. GSA officials said the USAi platform will also expand workforce readiness, with dashboards and analytics that let agencies track AI maturity and guide adoption strategies. The system allows agencies to evaluate different AI tools before integration, a step GSA leaders said will build trust in emerging technologies. Read more: US and China Chart Diverging Paths in Global AI Action Plans What Amazon, Meta, Uber, Anthropic and Others Want in the US AI Action Plan