Honor codes are a relic of earlier times, not designed to “last into the current environment,” one expert said.
Photo illustration by Justin Morrison/Inside Higher Ed | Ionut Ispas and Dmitrii Maslov/iStock/Getty Images
The already-small number of colleges with full-fledged, student-enforced honor codes is dwindling. After a three-year pilot of proctored exams, Stanford University student, faculty and administrative leaders decided in April that the university will allow proctoring for all in-person tests starting in September. Princeton University faculty approved a similar plan a month later. In making those decisions, both institutions grappled with students’ increasing use—sanctioned and not—of artificial intelligence.
AI is at the “forefront” of honor code reform, said JT Torres, director of the Houston H. Harte Center for Teaching and Learning at Washington and Lee University. The technology is everywhere—baked into browsers and word processors, embedded in learning management systems and in students’ phones, smart watches and sometimes even glasses. It’s ubiquitous and tempting, especially for young students under pressure, according to Tricia Bertram Gallant, director of the Academic Integrity Office and Triton Testing Center at the University of California, San Diego.
A typical honor code asks students to adhere to a set of academic integrity and behavioral standards that prohibit acts like cheating, plagiarism, lying and stealing, and enlists students as the enforcers of those standards. The code only works as well as the students who enforce it, and success relies on students’ desire to adhere to social norms, Bertram Gallant explained.
Established at many institutions in the early 20th century, honor codes were not designed to “last into the current environment,” Bertram Gallant said. “It’s a very, very unusual construct in 2026.” Experts agree: The codes aren’t working at their best. But whether they should be reformed—and, if so, how—is up for debate.
For students, honor codes foster trust, and the consequence for breaking that trust can be steep. “If you’re caught once, you’re expelled,” Torres said of students at Washington and Lee.
For institutions, an honor code sends a message to parents, employers, prospective students and others that says, “‘If you come here, you can trust your peers that they won’t cheat because we have this honor code. You can trust your peers that they won’t steal your backpack—you can leave it out and no one will touch it because of the honor code,’” Bertram Gallant said. “And then [institutions would say], ‘Our graduates are more trustworthy, not only because of the honor code, but because they chose a school with an honor code.’”
But AI is not a fellow student, and alone in a dorm room with ChatGPT, Claude or Gemini, the social norms that govern an honor code have waning influence.
“It’s different to say, ‘I agree not to look at my neighbor’s exam,’ because that involves another person and you’re thereby embroiling them in something … But AI is new, and people lack boundaries about when it’s appropriate [to use] and when it’s not. It’s just way harder to resist that type of cheating source than other types of cheating sources,” Bertram Gallant said.
Surveillance Is Just a Start
According to a 2025 Inside Higher Ed survey of more than 1,000 students, 85 percent had used generative AI to complete coursework. Over half—55 percent—said they used it for brainstorming ideas, 44 percent used it to edit or check their work, a quarter used AI to complete assignments or coding work, and 19 percent used it to write free responses or essays. What types of AI use constitute cheating vary by institution, professor, class and assignment, and that makes it difficult for students to parse what constitutes an honor code violation.
“Some have a hard time recollecting the extent to which the machine did things for them, because they’re involved—they prompt it, they get the output, they prompt it again,” Bertram Gallant said. “I’ve literally had students say, ‘But I typed everything myself. I didn’t copy and paste it.’ There’s this passivity, and yet, at the same time [they’re] fooling themselves that they actually created more of this than they did.”
Even if a student does violate their institution’s honor code through AI use, it’s not necessarily intentional, Torres said. It’s not as straightforward as looking at another student’s answer sheet or copying someone else’s essay. “I don’t really encounter students who are actively looking for a way to cheat. They’re looking for a way to succeed, and sometimes using AI, in their mind, is helping them get an edge,” said Torres, of Washington and Lee.
Genny Freed, a rising senior at the University of Virginia, is chair of the public university’s honor committee, which has frequently discussed how to respond to AI-enabled cheating, she told Inside Higher Ed in an email. Given the variations in professors’ AI policies, the committee is currently working to “encourage faculty members to clearly communicate their course’s AI policy to students,” she said.
“The Honor Committee does not use or admit AI detection tools, as the results are often unreliable. Instead, we evaluate the syntax and content of the assignment in comparison to the student’s prior writing or problem-solving methods and the material covered in the course,” Freed wrote. Most cases are resolved through what she called “informed retraction”—in other words, the accused student admits they are guilty before an investigation begins.
Introducing proctoring—and shifting honor code enforcement from the students to the faculty—so far seems to be the most common honor code reform. Bertram Gallant doesn’t believe schools should abandon their honor codes, but she does think increasing observation is a good idea.
“It is the professor’s job to closely observe students’ knowledge and skills, because they’re the ones verifying whether the students master—and to what extent the students master—the learning objectives or not,” she said. To have them proctor exams and say, “‘I know it’s this student who [cheated], and I know what aid, if any, they were using’—I just don’t see how that’s incongruent with an honor code.”
But surveillance can only work so well. What’s really needed is a change in how students are assessed, Torres said.
“We had an alumni board visit recently, and a couple of board members were sharing how, when they hire people, they essentially provide the candidate with a novel problem and messy data, and then they say, ‘Use whatever tools you want. How would you approach this?’ … so they’re not assessing what’s in your head, they’re assessing how you approach a problem,” he said. “We’re not even there in higher ed.”