The Medical University of South Carolina has released an AI Acceptable Use Framework for Academic Tasks, a five-category model that gives faculty a consistent way to communicate expectations about artificial intelligence in coursework. The framework, added to MUSC's Student Guidelines for Plagiarism and Artificial Intelligence, moves the conversation beyond cheating and toward AI literacy, transparency, and trust.
The framework emerged from two institutional AI strategic goals: pioneering changes in healthcare delivery, research, and education through AI, and creating an AI-competent workforce. Leaders at MUSC say the framework positions the university not just as an adopter of emerging technology but as a force in shaping how healthcare professionals learn to use AI ethically.
Five categories for AI use
The framework defines five levels of acceptable AI use in academic tasks. The first category, No AI, applies to exams and skills evaluations. AI planning allows students to use AI before beginning a task. AI limited permits AI for specific aspects, such as designing poster presentations or getting feedback. AI extensive applies to complex data analysis and clinical decision-making practice. AI exploration requires students to use AI to investigate its role in healthcare practice.
Every category except the first could require documentation of how AI was used. Students might submit a log of AI interactions or an explanation of their AI usage. For educators working with AI for Education, this documentation requirement offers a practical model for tracking student engagement with the technology.
"Students want clearer guidance about AI use," said E'lise Nissen, director of AI in Education and Scholarship at MUSC. "The framework gives faculty a consistent way to communicate expectations while helping students understand not only what is allowed, but why."
Built for health sciences education
Julaine Fowlin, executive director of the Center for the Advancement of Teaching and Learning at MUSC, said the framework was adapted from the Artificial Intelligence Assessment Scale, published in the Journal of University Teaching and Learning Practice in 2024. MUSC refined the scale to meet the needs of health sciences education, where critical thinking, clinical judgment, and patient safety are essential.
"The goal of education is to prepare students for authentic problem-solving in the real world," Fowlin said. "We should be asking: What core competencies must our learners achieve? What does evidence of that learning look like? And how does AI fit into that picture?"
The framework is one component of a broader AI toolkit at MUSC. All new students take a course called AI at MUSC, and employees have access to AI literacy training. Fowlin said this shared foundation ensures that transparency and responsible use are understood in context, not just mandated from above. Educators looking for structured approaches to classroom AI policies may find parallels in AI for Teachers training pathways.
Nissen said the framework addresses a trust gap. "Right now, many students do not trust higher education to guide them in when, how and why to use AI," she said. "I believe the Acceptable Use Framework is one of the most important tools we have for providing that guidance."
National attention and next steps
The framework has gained national attention, featured in webinars and discussions that are helping other institutions think about acceptable AI use in healthcare education. MUSC is developing a governance structure to ensure AI adoption continues while maintaining the ability to adapt as the field advances. The university is also working with the international Digital Education Council and the American Association of Colleges and Universities.
Fowlin acknowledged that not everyone embraces AI. "We have students who are scared of it," she said. In cases where AI use doesn't directly affect clinical practice, faculty members are encouraged to offer alternatives.
Why this matters for education professionals
MUSC's framework gives educators a concrete, transferable structure for defining acceptable AI use in coursework. Instead of binary yes-or-no policies, the five categories map AI use to specific learning objectives and task types. For faculty developing their own AI policies, the documentation requirement offers a mechanism for accountability without relying solely on detection tools. The framework's grounding in learning outcomes rather than adoption metrics is a model worth studying for any institution wrestling with AI policy in the classroom.
Your membership also unlocks:
