Overview:
The $2 million OpenAI deal will bring ChatGPT Edu to all four CU campuses, but critics claim the rollout bypassed shared governance.
The University of Colorado’s new partnership with OpenAI is actively reshaping how artificial intelligence will be used on its campuses, despite ongoing concerns from faculty, staff and students.
The debate over AI’s role in the classroom heated up in February, when CU announced a systemwide partnership with OpenAI to bring ChatGPT Edu to all four campuses and the system office. The CU system office will cover the first year’s licensing costs of $2 million for approximately 100,000 users.
Faculty and staff access at CU Boulder went live on March 31, but student access, which was originally scheduled for the spring, was postponed until mid-August at the Faculty Council’s request. The rollout has sparked a sharp backlash from some campus community members, who say the issue is not just AI itself but how the decision was made.
“That has been actually the driving force behind a lot of faculty and also student unhappiness with the deal,” said Lori Emerson, a professor of media studies and director of the Media Archaeology Lab at CU Boulder. “The fact that the president seems to have completely bypassed student and faculty governance structures.”
Emerson said faculty members, particularly those involved in teaching and research, should have been consulted before the university committed to such a sweeping agreement.
“January is the first time that I heard about it,” Emerson said. “Immediately, I was disgusted.”
University leaders have framed the deal as a way to provide equitable access to a tool many people on campus are already using. According to CU Boulder, more than 28,000 students, faculty and staff had already created public ChatGPT accounts using @colorado.edu email addresses before the agreement was signed. The university says ChatGPT Edu offers a more secure, institutionally managed alternative because CU data submitted through the platform is not used to train public OpenAI models.Â
However, opposition to the announcement has since expanded into a larger campaign of dissent. A public letter opposing the OpenAI agreement has drawn hundreds of signatures from CU Boulder faculty, staff, graduate students and undergraduates. The signers argue that the agreement was developed without meaningful consultation with subject-matter experts in AI, pedagogy and academic freedom.
The letter raises four central concerns: privacy, academic integrity, instructor misuse and the broader educational costs of normalizing corporate AI tools in the classroom. Signers argue that even if CU data is not used to train OpenAI’s public models, users are still being asked to trust a private company with sensitive academic data.
“I think that higher education and other institutions like this are headed in the wrong direction,” said CU Boulder student Jamie Destefano. “I understand AI is great for certain things, but it’s not beneficial for the well-being of most people.”
The dissent letter also argues that AI tools can weaken student learning when they are used to shortcut reading, writing and critical thinking. It warns that students may lose out on key intellectual skills if they come to rely on chatbots to complete assignments instead of doing the work themselves.
That concern is not theoretical for CU Boulder student Gia Scaglione, who has already witnessed instructors relying too heavily on AI in the classroom.
“I have encountered so many teachers who can’t even answer questions about their own assignments because they didn’t write the instructions,” Scaglione said. “And the students are the ones to point out errors.”
Even so, Scaglione said the new partnership itself may not dramatically change student behavior because AI is already widespread.
“I think, in general, AI contributes to the loss of critical thinking skills,” she said. “But I don’t think that that specific partnership will affect that because everyone’s already using it already on their own.”
That tension sits at the center of the debate. On one side, university leaders say ChatGPT Edu acknowledges an existing reality and gives students and employees a more secure platform than the public version. On the other, critics argue that official adoption by the university normalizes a tool that many believe is already undermining teaching and learning.
“To me, and I think a lot of other faculty members, and probably students as well, a university education is not for learning how to use corporate tools,” Emerson said. “They’re coming here to learn how to think and to learn the history of their discipline.”
The university, for its part, has not signaled any retreat from the broader initiative. CU Boulder’s Office of Information Technology says training, workshops and readiness resources will continue throughout the rollout, and more advanced features, including reasoning tools, Codex and API access, are expected to arrive later in 2026 as the system finalizes support requirements.
For now, the compromise is a delayed student launch, not a canceled one. That means the real test of the OpenAI partnership may come this fall, when students begin logging in at scale and CU’s debate over AI shifts from contract language and governance to the day-to-day reality of how, and whether, the technology belongs in the classroom.
“It might sound extreme, but I actually think that the integration of corporate AI tools throughout the university has the potential to leave a generation of students without the ability to critically read, write, and understand how knowledge is produced, and without the ability to produce knowledge themselves,” Emerson said. “And I can’t think of a more consequential thing than that, to be honest.”
