College AI Policies Vary by Professor, Leaving Students Confused

Students at Philadelphia universities are navigating a patchwork of artificial intelligence policies that can change from one classroom to the next, creating uncertainty about when tools like ChatGPT cross the line into academic dishonesty.

At Temple University, Drexel University, and the University of Pennsylvania, individual instructors hold authority over whether and how students may use generative AI in their courses. The decentralized approach means a student could be required to use ChatGPT for an assignment in one class while facing potential academic integrity violations for the same behavior in another.

Why it matters

As AI writing tools become standard in professional settings, universities face pressure to prepare students for workplaces where these technologies are commonplace. But inconsistent policies within the same institution can undermine learning objectives and leave students uncertain about ethical boundaries—particularly when some professors require AI use while others treat it as cheating.

How three universities handle AI

Temple introduced its generative AI guidance in 2023 after consulting hundreds of faculty members and more than 100 students. The university's Student Code of Conduct identifies unauthorized use of AI as potential cheating, but explicitly gives professors authority to set course-specific rules. When instructors don't state a position, Temple's default policy prohibits AI use.

"There is no university-wide policy on whether students may use generative AI in coursework," said Dana Dawson, assistant vice provost and senior director of Temple's Center for the Advancement of Teaching. "Policies are set by faculty members on a course-by-course basis because every course and every discipline is different."

Drexel's academic integrity policy similarly makes students responsible for understanding each instructor's written and verbal guidance on acceptable AI use. When AI is permitted, students must follow specific instructions about attribution and citation.

Penn takes a comparable approach but adds emphasis on privacy protections. The university distinguishes between Penn-licensed AI tools and publicly available systems, warning students against entering identifying information into unlicensed platforms. Penn provides access to several university-supported tools including ChatGPT EDU, PennChat, Grammarly Pro, and Microsoft Copilot.

Student experiences reveal confusion

Azaira Gonzalez, a Temple junior studying secondary education Spanish, described conflicting messages. "As an education student, we've always been told not to use AI, especially in lesson plans," she said. "So, for me, it's very conflicting when I'm hearing, 'Yes, use it. No, don't use it.'"

Alexa Lapiana, a Temple junior studying psychology, completed an assignment last semester that specifically required using ChatGPT to compare AI-generated responses with student research. She said the experience raised questions about boundaries, particularly when some professors use AI detection tools to identify potential violations.

Lapiana uses Grammarly for writing assistance and has sent multiple emails to professors this semester asking whether the tool could trigger AI detection flags. "I feel like the policies just need to be better identified," she said. "There just needs to be some more written explanation."

The concern highlights challenges universities face as AI becomes embedded in tools students have used for years. The distinction between writing assistance software and generative AI systems that produce original text is not always clear to students navigating course requirements.

Dawson said Temple's Center for the Advancement of Teaching encourages professors to state their expectations clearly in syllabi, include policies in assignment instructions, and remind students of course rules throughout the semester.

These details were first reported by Leah Duffy in The Philadelphia Tribune.