US Universities: AI Committees and Anti-Cheat Rules

Faced with the rapid adoption of AI on campuses, several American universities are structuring their governance, opening secure access to models, and clarifying current uses. Drexel, Boston, and Missouri detail committees, portals, evaluation policies, and training, with an equipment timeline extending to 2026.

Explicit Rules in Courses and Evaluation

At Drexel, teachers have considerable autonomy in using AI, provided they specify the permitted uses in their syllabi. The university committee has decided that AI detectors should not be used to establish cheating due to the risk of false positives. At the University of Missouri, teachers must also indicate in their course plans whether and how AI can be used, may use AI tools to assist with feedback but not to grade assignments, and AI detectors cannot serve as evidence of cheating. These measures frame evaluation while allowing for certain pedagogical uses.

Tool Access: Copilot, ChatGPT Edu, and a Multi-Model Portal

Drexel, which utilizes Microsoft technologies, offers Microsoft Copilot to teachers and staff and is gradually rolling out ChatGPT Edu licenses, with access planned for all students at the beginning of the fall 2026 term. In Boston, the AI Development Accelerator has established a unique portal providing access to ChatGPT, Microsoft products, Google Gemini, Claude, and Llama, allowing each user to choose the model that suits their needs. In Missouri, the Show-Me AI platform, designed for teaching and learning, enables the creation of class groups and personalized AI assistants, such as a writing coach tasked with providing feedback without writing for the students. The deployment of Show-Me AI among teachers and students is currently underway.

Cross-Functional Governance and Permanent Committees

Cross-functional structures are developing to avoid isolated management of AI. Jenay Robert from EDUCAUSE emphasizes that AI impacts strategy, policies, and operations, and effective governance must include all stakeholders. At Drexel, a task force convened in spring 2023 led to the creation of a permanent committee co-chaired by Steven Weber, who stresses the rapid evolution of AI and the diversity of responses across disciplines and funders. In Boston, two task forces, one academic and the other administrative, converged towards critical acceptance and evolved into AIDA, an advisory entity that develops guidelines and organizes events and training. John Byers, director of AIDA, highlights the importance of a community bringing together teachers, students, and staff, believing that responses to usage questions are more relevant when they come from peers. In Missouri, the desire for a secure environment for data and course materials led to the creation of Show-Me AI, with ongoing adjustments to features and associated training. Teachers collaborate with IT experts, a cooperation deemed valuable by Kevin Brown.

Pedagogy, Literacy, and Support

Drexel plans to integrate AI literacy into mandatory first-year courses to help students use these tools without losing their own voice, according to Steven Weber. In Boston, AIDA supports pedagogical innovation through guidelines and training while addressing common questions about syllabus mentions, managing AI-related plagiarism, usage in research, or employing data subject to regulatory constraints. In Missouri, training sessions support the adoption of Show-Me AI and its integration into classroom practices.

Adoption Pressures, Unregulated Uses, and Access Equity

The adoption of AI is advancing rapidly, sometimes faster than the establishment of governance frameworks, leading to experimentation outside official IT channels. This dynamic fuels unregulated uses and exacerbates disparities among students, with some having access to advanced models while others are limited to free versions. Teachers face challenges in maintaining academic integrity without becoming investigators. To address these issues, task forces have defined guidelines, updated integrity policies, centralized advice, and cataloged use cases, while initiatives like widespread access to tools aim to reduce equipment disparities.