Answer extracted from the 3 Takeaways™ podcast — listen to the full episode below.
What are Katrina Mulligan's three key takeaways on AI for government and national security leaders?
AI is no longer a future policy question but core infrastructure for public service and national power. Trust is the rate limiter—organizations must move beyond a defensive crouch on the trust question. The real opportunity lies in redesigning work itself, not simply adding a chatbot to old processes.
AI as Core Infrastructure, Not a Technology Question
Katrina Mulligan's first takeaway reframes how government leaders should think about artificial intelligence. AI has moved from the hypothetical to the operational—it is already shaping how nations compete, solve problems, and deliver public services. This is not a future scenario; it is the present reality facing every senior government official.
The speed of this shift is itself striking. As Mulligan explains in the episode, the timeline at frontier AI companies like OpenAI runs at an entirely different pace than government typically operates. A month at OpenAI feels like a quarter anywhere else—meaning the technological landscape government leaders must navigate is evolving faster than traditional policy cycles can accommodate.
Trust: The Real Bottleneck
Her second takeaway tackles the obstacle most organizations face but rarely name directly: trust is the rate limiter. Organizations spend enormous energy on defensive postures—risk mitigation, guardrails, and regulatory compliance—rather than building genuine confidence in AI systems and their outcomes.
Moving beyond a defensive crouch means shifting the conversation from "what can go wrong" to "what are we trying to accomplish." This requires a fundamental reorientation of how government agencies approach AI adoption, shifting focus from controls to capability and public benefit. The point is detailed in this podcast, where Mulligan's experience bridging both the Pentagon and OpenAI gives her unique insight into why this shift is so difficult—and so necessary.
Redesign Work, Don't Just Automate It
The third and perhaps most actionable takeaway addresses a common misstep: simply adding AI tools to existing workflows misses the transformative opportunity. True value comes from rethinking how work is structured, not from bolting a chatbot onto legacy processes.
This distinction matters enormously. ChatGPT deployed across Pennsylvania's state workforce, for instance, saved the average public servant 8 hours per week—a result that only emerged when agencies redesigned how their employees actually worked, not when they just gave them access to a new tool. Mulligan's own experience bears this out: she estimates she is at least 30% more effective than she was a year ago because of how she has matured her use of these tools, not simply because the tools exist.
For a deeper dive into how government should fundamentally restructure its approach to AI funding and deployment, listen to the full conversation where Mulligan discusses specific programs of record and the scale of change required.
"I honestly think that I am at least 30% more effective, maybe more than I was a year ago, because of how I've matured my use of these tools. I'm a better leader. I'm a better manager."
Katrina Mulligan — Head of National Security Partnerships at OpenAI, former senior official at the Department of Defense, National Security Council, and Department of Justice, where she served as number two at the Pentagon overseeing special operations. Her unique position bridging both Washington national security structures and frontier AI development gives her rare authority on how government must adapt to AI's accelerating pace.
See also
How should government change the way it funds and thinks about AI to unlock its full potential for public services?
Mulligan argues that government must stop funding AI purely as an IT line item and begin allocating resources to real, sizable programs of record targeting measurable outcomes in public service delivery.
How far ahead is the United States from China in AI development, and is that lead secure?
Mulligan estimates the U.S. has opened up its lead to a single-digit number of months — approximately six to eight months ahead, up from roughly four to six months previously.
How does China's approach to AI differ from the United States, and could America build the best AI and still lose?
Mulligan identifies three key differences: China has much higher public trust in AI, the Chinese government and CCP have a more direct seat at the frontier of AI development, and the integration of AI into governance structures is far more comprehensive.
Key takeaways
- AI is now core infrastructure for national power and public service delivery—not a future concern but a present reality shaping how governments compete and solve problems.
- Trust is the rate limiter for AI adoption; organizations must move beyond defensive risk postures and focus on building genuine confidence in AI's outcomes and benefits.
- Effective AI deployment requires redesigning workflows, not simply adding tools; real gains come from rethinking how work is structured, as demonstrated by Pennsylvania's experience saving public servants 8 hours per week.
- Government must fundamentally recalibrate its understanding of timeline and pace; a month at frontier AI companies feels like a quarter in traditional organizations, demanding new approaches to policy and funding.