5 Takeaways from the AI Summit

The Pentagon’s rollout of GenAI.mil shows how secure, enterprise-wide AI access can rapidly shift organizational culture, moving from basic workflow automation to applying AI at scale across science, engineering and weapons development.

War Department officials see AI as a direct mission advantage, accelerating progress in areas like materials science while building user confidence that AI can help DOW outpace adversaries.

“We’re starting to see the implementation of, ‘Well, I can do this with AI in my enterprise level stuff, but what can I do for a scale type of science now?’ And then you can boil that down to material science or seeker technology or RF, you can go on and on,” said Jake Glassman, deputy assistant secretary of war for science and technology foundations.

The Trump administration’s July AI Action Plan calls on National Institute of Standards and Technology (NIST) to update its AI Risk Management Framework. Officials say the revision will reflect a government-wide shift toward value-driven AI, easing regulatory burdens so agencies can deploy use cases that deliver returns commensurate with their investment.

Officials emphasized that trust remains the linchpin. AI systems must be reliable and transparent enough to be used at speed, or risk management itself becomes a barrier to mission impact.

“It’s very important that people trust their AI, or else they won’t use it. And then that investment becomes not a good investment,” said NIST AI and Cybersecurity Researcher Martin Stanley.

The General Services Administration is focusing on making AI tools accessible to non-technical users through platforms like USAi.gov as federal agencies continue to experiment with the tech.

The next phase of federal AI will depend on hands-off tools that embed AI into day-to-day operations, GSA Chief AI and Data Scientist Zach Whitman said.

“I think the next step is going to be on the hands-off type of direct interaction with non-technical folks, rather than it being for just the developers, or the engineers, or the people in the business,” Whitman added. “Making this technology approachable in a hands-off way for non-technical folks in a safe and observable, controllable way for enterprise is, to me, where I expect this year and next year to go.”

President Donald Trump’s November executive order launching the Genesis Mission signals a shift toward building the data, compute and automation backbone needed for global AI superiority.

“If we’re going to move from decades to days, you need autonomous capability, and it needs to be assured,” said Pacific Northwest National Laboratory’s (PNNL) Center for Continuum Computing Director Scott Godwin.

By prioritizing clean data, autonomous experimentation and shared platforms like the American Science Cloud, federal leaders are focusing less on isolated use cases and more on creating scalable, secure infrastructure that turns federally funded research into sustained strategic advantage.

As federal agencies face workforce reductions and growing data demands, IT officials say AI-powered networking is essential to maintain operations.

“Reduction in force … has really taken a toll,” said Verizon’s Federal Senior Systems Engineer William Shealy, pushing agencies to rely on AI-driven routing, agentic models and secure interconnects that can “look at it, think about it and do it” without constant human intervention.

He warned that as systems become autonomous, humans must maintain responsibility for AI outcomes. “It’s a look before you leap type of thing … just because we could, doesn’t mean we should,” Shealy said.