Unfixed Newsletter — May 14–June 10, 2026
The AI Higher Ed Breakdown
Editors’ note
Just as a reminder we are on a 4-week cycle during the summer and will return to a 2-week cycle when the academic year starts.
The past few weeks were about AI and shifting institutional relationships. Campuses are signing systemwide contracts, students are using AI unevenly across majors, career offices are being rebuilt around labor-market anxiety, and search is turning into an agent layer. Even the companies selling these systems are changing shape as OpenAI, Anthropic, and SpaceX move toward public markets. All of these stories have trickle down effects into University boardrooms and classrooms.
The stories
1) Campuswide AI contracts are becoming institutional defaults
The California State University system renewed its ChatGPT Edu contract after an initial systemwide rollout that drew both enthusiasm and criticism from students, faculty, and system observers. CSU’s own FAQ says the renewal followed a systemwide evaluation process and a recommendation from its GenAI Advisory Committee. At the same time, the University of Chicago announced a partnership with Anthropic that will provide Claude Enterprise to faculty and staff in July and to students before the fall term.
Why this matters: These are procurement stories, but faculty will feel them as teaching stories. Once an institution buys a systemwide AI platform, the tool becomes part of the instructional environment whether the instructor wants it there or not. That changes what students expect, what faculty have to explain in syllabi, and what campuses need to answer about data handling, opt-outs, intellectual property, accessibility, vendor lock-in, and contract continuity.
Sources: https://www.calstate.edu/impact-of-the-csu/technology/ai-empowered-csu/Pages/ai-faq.aspx
https://localnewsmatters.org/2026/05/27/csu-renews-openai-chatgpt-contract/
https://www.calmatters.org/education/2026/05/california-state-university-open-ai-chatgpt-contract/
https://president.uchicago.edu/From-the-President/Announcements/AI-Tools-at-UChicago
2) The largest student AI-use study so far points back to assessment design
A new study in Science, covered by Cornell and UC Berkeley, surveyed more than 95,000 students across 20 U.S. public research universities. About 37 percent of students reported using generative AI at least monthly for coursework, and about 9 percent reported using it in ways the researchers classified as cheating. The rates varied sharply by field and student group, with computer science students using AI far more often than arts students and with notable gaps by gender, income, and race.
Why this matters: This is a cross-institutional study which helps us understand the whole higher ed landscape. This study also calls into question early responses to AI in the classroom. For instance, a ban does not produce equal conditions if some students are already using AI fluently and others are not. A permissive policy does not solve the problem if assignments still depend on final products that no longer show how the work happened. Faculty need assessment designs that make process, judgment, source selection, revision decisions, and explanation visible. They also need discipline-level norms, because the AI question looks different in computer science, writing, art, business, engineering, and teacher education.
https://www.science.org/doi/10.1126/science.aec5115
3) Dartmouth puts career design at the center of the AI jobs conversation
CNBC’s May 18 story used Dartmouth as a case study in how colleges are responding to student anxiety about AI and employment. The underlying investment is substantial: Dartmouth says its Center for Career Design is backed by a $94 million fundraising goal, with more than $61 million already raised, and includes one-on-one coaching, custom AI-powered career tools, expanded internship funding, and access to its alumni network. Earlier this year, Dartmouth announced $30 million in endowed gifts to expand internship support, including funding for unpaid or underpaid internships.
Why this matters: This is the other side of the AI-and-jobs debate. Colleges cannot answer student anxiety with broad claims about the enduring value of a degree. They need concrete pathways into work, especially if entry-level white-collar roles become more competitive, more automated, or more dependent on AI fluency. Dartmouth’s response is expensive and elite, so it is not a simple model for everyone else. Still, it captures the direction of pressure: career services, advising, internships, alumni networks, and curriculum are going to be judged together. Faculty will increasingly be asked to explain how courses build the judgment, adaptability, communication, and domain knowledge students need when the first rung of the job ladder is less stable.
https://home.dartmouth.edu/news/2026/04/career-center-built-lifetime-fulfilling-work
https://home.dartmouth.edu/news/2026/01/dartmouth-raises-30-million-offer-internships-all
4) Google Search is turning into an agent layer
At Google I/O, Google described “a new era for AI Search.” The most important part was not another summary box. Google is pushing Search toward AI Mode, multimodal interaction, task-monitoring agents, booking tools, automated calling, custom generative interfaces, and connections to personal apps such as Gmail, Calendar, and Photos. In Google’s framing, Search is becoming a place where users ask a question, track a task, compare options, and delegate action.
Why this matters: There are layers to the implications of this story. First, students and faculty do not have to choose a chatbot for AI to reshape their work. If search itself summarizes, filters, personalizes, and acts, then ordinary research behavior changes by default. This makes “banning” AI almost impossible if you also want students to do research. Second, in changes the information literacy conversion since each search comes with context. Third, a surprising amount of work in multiple fields was built around optimizing for search engines, specifically Google. That landscape is sure to change and the question of how to prepare students for it is an open one.
Source: https://blog.google/products-and-platforms/products/search/search-io-2026/
5) AI companies are moving toward Wall Street pressure
The AI IPO wave became more concrete this week. Anthropic announced on June 1 that it had confidentially submitted a draft S-1 registration statement to the SEC for a proposed IPO. On June 8, Axios reported that OpenAI had also confidentially filed draft IPO paperwork. Axios previously reported that SpaceX had filed for an IPO as well, making this a broader public-market turn for companies tied to AI infrastructure, models, and platform power.
Why this matters: Higher education is making long-term teaching, research, and administrative decisions around vendors whose economics are still unstable. Public markets reward growth, retention, pricing power, and product expansion. Campuses should expect more pressure toward enterprise contracts, bundled services, usage targets, premium features, workflow lock-in, and data-connected tools. That does not mean universities should avoid these systems altogether. It does mean that faculty and administrators should stop treating current pricing, current privacy terms, and current product boundaries as permanent. A tool that feels optional in 2026 can become infrastructure quickly if a campus builds enough work around it.
Sources: https://www.anthropic.com/news/confidential-draft-s1-sechttps://www.axios.com/2026/06/08/openai-ipo
https://www.axios.com/2026/05/20/openai-ipo-spacex-musk
https://www.axios.com/2026/05/20/elon-musk-spacex-ipo
6) Wildcard: An AI ethics column became an AI disclosure problem
The Sydney Morning Herald removed an opinion piece by Western Sydney University professor Cath Ellis after learning that she had used generative AI to help write it. The piece had warned students against using AI to cut corners. Ellis reportedly used Microsoft Copilot and uploaded a large sample of her own scholarly work to help generate the column. The problem was not simply that AI was used. The issue was that the use was not disclosed under the publication’s policy.
Why this matters: This is the kind of strange story that gets passed around because of the apparent hypocrisy. The more useful lesson is about norms. Faculty are often asking students to disclose AI use, explain process, and distinguish assistance from substitution. Those same questions now apply to faculty writing, public scholarship, peer review, grant writing, committee work, and institutional communication. A coherent AI policy cannot only police students. It has to define what disclosure means across the institution and what kinds of assistance are compatible with authorship, accountability, and professional judgment.
Sources: https://www.abc.net.au/news/2026-06-03/sydney-morning-herald-takes-down-ai-wsu-article/106756298
From Our Work
CSU, ChatGPT Edu, and the limits of cancellation
It was a big month for us and we are grouping a couple things together. Nik and Zach wrote about why canceling CSU’s ChatGPT contract would not solve the AI problem, and EdSource published the larger news story about the contract in which we are quoted. Zach was also quoted in NPR’s coverage of CSU’s AI rollout, and Nik was featured in the New York Times Magazine piece on the same topic.
Links:
New York Times Magazine article
Unfixed Episode 30: Preparing Teachers for the Age of AI
Nik and Zach talked with Jamie Gunderson about teacher preparation, AI literacy, and what future educators need before they walk into classrooms where students and institutions are already using these tools.
Link: https://www.meltsintoair.org/unfixedpodcast/preparing-teachers-for-the-age-of-ai
Unfixed Episode 31 [Special]: CSU Renews Its OpenAI Deal: Zach and Nik React
The California State University system just renewed its OpenAI contract, turning what began as a bold AI pilot into a long-term institutional commitment. In this special Unfixed reaction episode, Zach and Nik break down the costs, controversies, and consequences of the ChatGPT Edu deal—from academic freedom and governance to vendor lock-in and the future of teaching. Is CSU building essential educational infrastructure, or making a risky bet on a single AI company?