The university that hosts its own AI: BuddyGPT bets on data sovereignty at UCO's campus

By News 9 · October 12, 2026.

The University of Central Oklahoma (UCO) has launched BuddyGPT, an artificial intelligence platform that runs on the university's own servers. According to News 9, it is available free of charge to students and faculty, including for research computing tasks.

According to professor Brad Paynter, the tool answers questions and generates charts. It can also organize and search large document collections and answer questions about their content. In other words, a conversational assistant capable of working with in-house material.

The central argument is privacy. Emily Lyons, associate professor of Mathematics and associate dean, says that because it is local, "intellectual property stays within the institution." The university says it will continue developing the platform as it upgrades its supercomputer and as the technology evolves. The article does not specify which model underlies BuddyGPT, how many users it has or what it costs to operate, so we cannot assess its quality against commercial alternatives.

Our take: what matters is not BuddyGPT itself but the pattern. When a researcher pastes unpublished data into an external service, they give up control over something valuable. Hosting the model in-house removes that risk and turns privacy into an architectural decision rather than a matter of trust. Broadly, this is possible because the open-model frontier has advanced quickly and now allows a mid-sized institution to run its own assistant on hardware it already had or is expanding.

There is an honest caveat: local does not always match the best of the closed market, and maintaining it requires staff, energy and updates. For a university, the cost may pay off if usage is heavy; for others, perhaps not.

We also think the teaching approach is sound. Paynter does not ban AI but teaches students to use it responsibly, and Lyons stresses cross-checking sources and verifying answers rather than taking the output on trust. The university's industry partners, according to Lyons, are asking for graduates who master these tools. In the short term, that labor-market pressure is real and will redefine what gets assessed in the classroom. In the long term, a generation trained in critical thinking alongside AI is exactly what a society needs if it aspires to delegate routine work and devote itself to what it is passionate about.

The thesis: democratizing AI is not just about cheaper models, but about institutions that run them with sound judgment and under their own rules.