Colorado Students Push Back as Universities Sign Million-Dollar AI Deals

The University of Colorado system committed $2 million to an OpenAI partnership in the last academic year, rolling out a CU-branded version of ChatGPT to its campus. By March 2026, more than 28,000 people at CU Boulder had university-linked ChatGPT accounts. But a visible group of students is pushing back hard, calling the tool a "plagiarism machine," raising data privacy concerns, and questioning whether a university's job is to prepare students for corporate AI workflows at all.

What happened

Data point	Detail
CU system AI deal value	$2 million partnership with OpenAI
CU Boulder ChatGPT accounts (by March 2026)	More than 28,000 university-linked accounts
Faculty and staff accounts	More than 3,000
CU Boulder total enrollment	Nearly 39,000 students
Colorado State University	Separate deal to create its own CSU version of ChatGPT

During the last academic year, the University of Colorado system announced a $2 million deal with OpenAI to give students and faculty access to a university-specific ChatGPT. Colorado State University in Fort Collins followed with its own separate agreement. CU pitched the move as an equity play: if tens of thousands of students are already using these tools, the university said it would rather teach responsible use than pretend the technology does not exist.

CU Vice President of Communications Michele Ames framed it plainly. Preparing students to work with AI, she said, is doing right by them. Ignoring the technology and the workplaces they will enter would be “doing them a disservice.”

Who is resisting, and why?

Mirakle Wright, a 25-year-old graduate student at CU Boulder, noticed something the broader AI adoption narrative was missing. Many of her peers actively avoid generative AI, and there was almost no research on them. She began documenting their reasons in a small study of self-described AI-avoidant students.

The objections she recorded fell into several categories:

• Environmental concerns, particularly around water use in AI data centres
• Data privacy worries about what gets shared with OpenAI
• Objections to AI training on copyrighted creative work
• Personal experience of what students called “cognitive decline” while using the tools
• A philosophical rejection: universities should foster critical thinking, and AI does the opposite

Wright compared the holdouts to wartime conscientious objectors. “People have such negative feelings about AI. They’re becoming resisters,” she said.

CU information science professor Casey Fiesler, who studies AI ethics and advises Wright, said she has watched student sentiment shift. Among some younger users, generative AI is now considered “cringe.” She has seen students call each other out: “Seriously, dude, you can’t write an email on your own?”

The loudest voice in the room

Communications student Nate Wasylik took his objection directly to the CU Board of Regents at an April meeting, asking the governing body to cancel the OpenAI contract. “I am horrified at this board’s handling of the OpenAI contract,” he told them. In a later interview, he described ChatGPT as “the giant plagiarism machine” and said he has no interest in measuring his education against a “billionaire’s mindset.”

Elizabeth McKinnie, a 28-year-old Ph.D. student in CU Boulder’s information science department, brings a sharper technical perspective. She was formerly a software engineer at Microsoft, where she had early access to ChatGPT. The experience was enough to make her want to leave software engineering entirely. On February 10, she attended a talk by Timnit Gebru, former co-lead of Google’s ethical AI research team, about the harms of generative AI. Hours later, she received the university email announcing the OpenAI partnership.

Why it matters

This is not a niche campus debate. Universities are among the largest institutional buyers of AI tooling right now, and the CU situation is a preview of the friction that follows when institutions move fast on enterprise AI contracts without a clear student-consent framework. Data privacy is the sharpest edge: when a university links student accounts to a commercial AI provider, it is hard to know exactly what data flows where and under what terms.

For businesses watching this, the same dynamic plays out in workplaces rolling out AI tools to reluctant employees. The objections are similar: loss of cognitive ownership, distrust of the vendor, and a feeling that the tool is being imposed rather than chosen. Mandating adoption rarely works as well as giving people a genuine reason to use it. Our coverage of how one MBA student built a personal AI productivity system shows what adoption looks like when it is voluntary and specific.

Our take

The universities are not wrong that AI literacy matters for graduates entering the workforce. But dropping a $2 million OpenAI contract on 39,000 students and calling it equity is a stretch. Equitable access to a tool students did not ask for is still compulsion with better branding.

The students raising data privacy concerns have a real point. Enterprise ChatGPT agreements vary widely in what they protect. A university signing on behalf of students is not the same as a student choosing to sign up. The “we’ll teach responsible use” argument works only if the university can actually audit how the tool is used and what data leaves campus, which most cannot do in practice.

For businesses thinking about AI integration in their own operations, the lesson here is process before rollout. Telling people they must use a tool, for their own good, is how you create a resistance movement. Showing specific, time-saving use cases and letting people opt into genuine workflows is a different story.

What to do about it

1. Review any enterprise AI contract your organisation has signed and confirm what data is retained by the vendor and for how long.
2. Separate “access to AI tools” from “mandatory AI use.” Giving people access is fine; requiring it without training breeds resentment.
3. Identify one or two concrete tasks where AI saves measurable time before asking anyone to change their workflow.
4. Watch how student-facing AI policies evolve at universities. They tend to preview what regulated industries adopt two or three years later.

The loudest adoption numbers mean little if the people behind them are looking for the exit.

Source: Bing News · OpenAI