When North Carolina State Treasurer Brad Briner announced a 12-week artificial intelligence pilot with OpenAI, he didn’t mince words: “This is about making every dollar work harder for the taxpayer.” That mission has taken a transformative turn with the launch of the first AI-powered initiative in the state government—a partnership that could signal a seismic shift in how public services operate in the U.S.
The pilot, launched by the North Carolina Department of State Treasurer in partnership with OpenAI, aims to harness the power of a customized ChatGPT model to perform bureaucratic heavy-lifting—summarizing lengthy reports, detecting red flags in audits, and combing through data to locate unclaimed property for residents. The best part? The state isn’t spending a dime to run it. In fact, the program runs on public data only, with strict adherence to privacy protocols and no personal information being shared. Source.
AI in the Public Sector: More than Just a Trend
AI applications in the private sector—from customer service chatbots to predictive analytics—are well-established. But the public sector remains relatively cautious. That’s what makes this North Carolina pilot one to watch. Unlike the public-facing version of ChatGPT, the tool being used here is a tailored model, fine-tuned with Treasury-specific scenarios and terminology to improve relevance and precision. OpenAI’s custom models initiative enables this high level of granularity.
So, what’s it actually doing? In practical terms:
- ChatGPT scans decades of financial audits to flag unusual spending patterns or late reporting that might otherwise be missed in manual reviews.
- The model compiles summaries of complex policy or performance reports that typically take hours to parse—saving employees valuable time and reducing human error.
- It helps staff identify unclaimed financial assets faster by analyzing huge datasets the way an entire team might—but in seconds.
According to a detailed release by the Office of the State Treasurer, the goal is to improve efficiency without cutting jobs. In fact, Briner emphasizes that the intention is to free up staff to focus on higher-impact tasks that require human judgment and empathy.
“AI is doing the digging,” he said. “People are making the decisions.”
OpenAI’s Government Footprint Expands
North Carolina is not alone in tapping AI for smarter governance. Pennsylvania and Minnesota have both run similar pilots, hinting at a growing trend. But North Carolina’s project is one of the most advanced in terms of administrative depth and the scope of its applications.
Aaron “Ronnie” Chatterji, OpenAI’s first chief economist, noted in recent press coverage that public-sector AI adoption holds “massive potential to reallocate time, reduce inefficiencies, and improve transparency.”
And here’s a surprising stat: According to Treasurer Briner, nearly $1 billion in unclaimed property is sitting in state coffers. If AI can recover even a small percentage of that more efficiently, the return on investment could be enormous—in dollars and public trust.
Where Privacy Meets Productivity
Of course, deploying AI in government raises fair concerns around data use and transparency. The North Carolina team has made data integrity a top priority. The system uses only public, structured datasets. Personal information is excluded, ensuring compliance with state privacy laws. As covered by WFMY News 2, the information processed involves existing public audits, budget documents, and publicly available unclaimed property data.
By limiting AI access to transparent and auditable inputs, the state is establishing a replicable framework that other agencies may look to adopt. Already, officials from other departments have expressed interest, as noted in interviews by WCNC Charlotte.
What’s Next? A Smarter Statehouse?
It’s not hard to imagine broader implications. Success here could lead to AI integration in departments handling everything from transportation to Medicaid. By allowing AI to take on pattern recognition, document synthesis, and proactive alerts, human workers can concentrate on policymaking and citizen support.
But this shift also raises long-term questions:
- Will job roles in government evolve into hybrid analyst-technologists?
- What upskilling will workers need to collaborate effectively with AI?
These questions are already in motion as this pilot—and others—attempt to define what an AI-augmented civil service looks like.
At its core, this initiative is about doing more with what you already have: data, talent, and time. If North Carolina’s pilot proves successful, don’t be surprised if this becomes the new playbook for government efficiency nationwide.
Conclusion
If an AI can sift through decades of financial data in seconds, what else have we been doing the hard way for too long? North Carolina’s pilot isn’t just about speeding up paperwork—it challenges the very model of how government can serve its people. Instead of replacing human decision-makers, it’s shifting the foundation beneath them—freeing up time, energy, and focus for what machines can’t do: exercising judgment, building trust, and solving complex societal problems.
This moment signals more than a tech upgrade; it’s the start of a philosophical shift in public service. When the tools of tomorrow land inside the halls of government, will we be ready to use them not just efficiently, but wisely? The experiment unfolding in North Carolina dares us to ask not what AI can do for government, but what a government empowered by AI might do better—for everyone.
