Updated October 16, 2026. Originally published January 20, 2024.
This update reviews public ASU and OpenAI documentation. GPTOnline did not access an ASU account or test a classroom rollout.
Arizona State University announced its collaboration with OpenAI on January 18, 2024. The university’s announcement described ASU as the first higher education institution to collaborate with the company and introduced ChatGPT Enterprise for faculty and staff exploration. The agreement began a university-led rollout whose access arrangements subsequently changed.
Cover image: Historical conceptual illustration retained from the original article. The robot, graduation outfit and campus scene do not document an actual student, campus event or product test.
The 2024 launch
The initial plan invited faculty and staff to propose uses for the technology through a challenge beginning in February. ASU identified student success, research and university operations as priorities. These were the initiative’s intended areas of work; the announcement did not establish measured learning gains.
A March 25, 2024 update confirmed that the first challenge round had launched on February 1. The second round expanded proposal eligibility to student researchers and included classroom projects. Accepted proposals received licenses, making individual participation dependent on the rollout rather than automatic access for every student at launch.
For background on the organizational product involved, see our ChatGPT Enterprise launch overview.
From Enterprise to Edu
OpenAI introduced ChatGPT Edu on May 30, 2024 as an offering for universities. Its announcement included an ASU language-learning project being developed to give students German conversation practice and feedback. That example described a particular teaching application; it was not evidence of a university-wide improvement in academic results.
On September 15, 2025, ASU announced a broader expansion, making ChatGPT Edu available at no cost to individual students, faculty, researchers and staff, with licenses beginning October 1. This later expansion is distinct from the proposal-based launch in early 2024.
Access today
ASU’s current ChatGPT Edu guide lists no-cost access for students, faculty and staff. Most users request it through the university’s Self-Subscription Portal, accept a workspace invitation and sign in with ASURITE credentials and DUO authentication. The ASU logo identifies the university workspace.
Student eligibility depends on active enrollment; employee access ends when ASU employment ends. Feature access also varies by account level. Student upgrades require faculty endorsement, while employee upgrades require a request with justification. Check the guide for current conditions before planning coursework around a feature.
The same guide says applicable data policies still apply. An institutional account does not remove the need to check whether particular course materials, research records or student information may be used.
Plan a small pilot
The following is a fictional planning example, not a report of an ASU project or a GPTOnline product test.
Imagine a university writing center considering AI-generated practice questions for a workshop. Its first decision should be what participants need to learn. A workable objective might be helping students identify unsupported claims in a short argument and explain what evidence would strengthen it.
- Define the activity. Ask the tool to draft practice passages containing a clearly specified weakness. Have an instructor check every passage and answer key before students see them.
- Choose permitted material. Start with invented passages written for the exercise. Keep student submissions, grades and personal records out of the pilot unless the institution has approved that use.
- Confirm access. Identify which workspace and features the exercise needs. Check every participant’s eligibility and provide an equivalent activity for someone who cannot use the tool.
- Set the classroom rules. Explain which AI assistance is allowed, what students must disclose and which parts of the task they must complete themselves. Include a way to question a misleading answer.
- Measure the right outcome. Use a short independent exercise before and after the workshop. Score students’ explanations with the same rubric rather than counting prompts or treating polished output as understanding.
- Decide what happens next. Record instructor correction time, access problems and recurring mistakes. Agree in advance what evidence would justify revising, expanding or stopping the activity.
Even an improved post-workshop score would need careful interpretation: students might improve through ordinary practice or instructor feedback. A larger evaluation could compare the AI-assisted activity with another teaching approach and check whether the skills persist. The pilot’s value is in producing evidence for a decision.
What to take away
The useful questions for another institution are practical: who receives access, who approves the material, what students are expected to learn and how the result will be evaluated. A partnership announcement establishes a starting point. A defensible teaching decision requires a clear activity, workable access and evidence gathered in the relevant setting.