AI Cheating, FERPA & Policy Design: A Q&A with Christian Moriarty (Part 2)

AI, the Courts, and Academic Policy — Part 2

A Q&A with Christian Moriarty on AI, academic integrity, and best practices for AI policy.

Christian Moriarty is an ethicist and attorney, director and treasurer of the International Center for Academic Integrity, and a professor at St. Petersburg College. He now spends about a week a month helping institutions build AI policy. His answers below are in his own words; the questions have been condensed, and his responses lightly cleaned only for transcription artifacts. The views expressed are Christian Moriarty’s own and do not constitute legal advice or necessarily reflect the views of Proctorio. Christian received no compensation for this interview and nothing said herein constitutes legal advice or is an endorsement of any product, service, or vendor, including Proctorio. This interview is published with Christian Moriarty's permission for use on the Proctorio blog only. It may not be reproduced, excerpted, quoted, or repurposed in any other context without his prior written consent — contact marketing@proctorio.com before reuse.

This is Part 2 of 2. Read Part 1 here.

You've said before that students are being actively incentivized to cheat. What do you mean?

Students are, as of right now, actively being incentivized to cognitively offload. The way we set up our assessments, particularly with asynchronous courses - it's in their interest to cheat. And if it's actively in their interest to do so, then of course they're going to. I'm at the position where I'm not even sure I blame them. I still get upset when they lie to my face about it, but I don't blame them entirely. You probably saw the Princeton story where they plan to begin in-person proctoring, for the first time ever: a bunch of students were upset, rightly so, that the only way they could compete was to also cheat. And that's not right or fair. So we've got to figure out where we're going with this, and how to align our incentives where it's easier to be honest than it is to cheat.

The thing that really radicalized me on this is hearing a student tell me that they felt obligated to purposely write worse in order to ‘not get caught by AI checkers.’ Just the combination of emotions that I felt by that; I was upset, I was mad, I was sad. This was a face-to-face student, and most of my work face-to-face is - they do it in front of me, so I take them at their word. This student genuinely wanted to learn and write for themselves, and I found them to be a good writer. But they felt obligated to dumb themselves down. And that…is not what we are here for. I don't want an honest student to have to game their writing to not be displayed as dishonest. That's untenable.

What would you recommend institutions do on this point?

You have to set expectations, but you also have to enforce expectations. Some institutions only have the second one. You have to coach and inspire to integrity, but you also have to hold them to those standards. The line I keep repeating over and over again is: the strongest correlation predicting whether a student is going to cheat is whether they think their peers are cheating. Not whether they actually are, but whether they think they are. And if you have a culture of, ‘I think everybody's going to cheat,’ that makes each individual much more likely to cheat. If instead you have a culture that is constantly reminding folks that cheating might feel easily accessible, but at this institution we don't tolerate it, and these are our expectations, you reduce the incidence of cheating, because people think fewer people are cheating.

Back to your legal perspective - Are there privacy landmines schools don't see coming?

A quarter of the time I'm just talking about the FERPA baseline: what are our privacy obligations to our students? You cannot put your student papers into a general purpose consumer AI tool.- You're breaching their privacy. Technically the punishment for breaching student privacy is that you lose all government funding, and while no institution has had that happen to them before, would you like to be the test case? Because if you lose government funding, you close, because that means loans too, not just tuition assistance.

What about detection tools? Isn't there something that just tells you, 100%, this is cheating?

People are looking for that, sure, and I think we all got too used to the view of pre-AI technologies that could more easily say, ‘this is a silver-bullet proof that you cheated.’ We had that for 20 years, and we need to train people out of that. There's not a silver bullet solution to any of this. It's going to be a combination of policies and practices, and we need to be more knowledgeable about how those align.

People come up to me after a talk and say, "What I do is I give a student paper to five different AIs, and if three out of the five tell me it's cheating, then I say it's cheating, and I don't give them a chance to respond. Is that okay to do?" There are at least two ethics issues and three legal issues why you should not be doing that, and we need to train practitioners to think differently so they already know the answer to a question like this.

The literature actually says that if your data shows cheating rates going down - that might be bad, because your students are cheating, you're just not catching them, or you're not doing anything about it. So take heart - when numbers go up, that usually means you're in a better place. It's counterintuitive, but it seems to be true.

Let’s say a school knows its policy is weak and wants to build a real one. Where do they start?

The development of good policy is hard, and it takes time, and particularly with artificial intelligence it's constantly changing, so it's inevitably going to have to be iterated on.

The best way to do it is to start with some philosophy of what the policy is, and then break down into levels of policy: faculty-facing, student-facing, etc, that can be iterated on or granted to individual faculty members for their use cases. But further from that, it's also important to do the ‘road show’ with it, to get the stakeholders involved. Making good policy is honestly the easier step. You have to then sell the policy. When we first made our college-wide academic integrity policy, in 2019, we had to get everybody on board with it and get everybody to start implementing it, and that took a whole year. But because we did the full road show, we went to every single department, every single student group, every administrator, and we got the buy-in we needed, and it stuck.

As far as stakeholders, you need an administrator, you need faculty members, you need students. You want their perspective on what pressure points they experience - and what they think is fair. Ultimately, you do need a lawyer in the room, but also don't let them run the room (and that's coming from a lawyer). You want to avoid your policies being overly legalistic as much as possible.

Two of my colleagues, Courtney Cullen and Greer Murphy, wrote this great research paper*. They did an evaluation of a hundred different academic integrity policies, and found that a large percentage had not been updated since before generative AI became publicly available. , aFurther, the average required grade level for comprehension of those policies was around 14th grade. Many of them were significantly legalistic, obviously written by lawyers as a means of protecting against litigation, and not with a view to being understandable. You can have student-facing versions and faculty-facing versions and make it understandable to the average person. Expecting an 18-year-old starting college to understand what a JD wrote is way too high of an expectation. It's not because your students are dumb, it's not because they don't care, it's because it's literally impenetrable. So getting it to a point where you can wordsmith it to where it's understandable, and once it's understandable, it's much more likely to be followed. Which is ultimately the goal.

So that’s the long term solution. What's one, simple thing someone can do tomorrow?

I think people undervalue just…talking about it with colleagues and with students. At least at my institution (and I suspect this is the case with others) we have precious few times where we just talk to each other anymore. Pre-COVID there was the lounge, the lunchroom, and we had more time to just commune, but now we're more and more offline, less and less on campus, and we don't have the opportunity for informal interactions anymore. We need to talk to each other about what we're seeing and any solutions that we have, and share resources.

As far as talking to students, that’s easier - I’m able to constantly talk to my students about this. When ChatGPT first came out, I asked how many of my students had even heard of it, and not a single student did. A year after it came out, they've all heard of it. Then I ask them, do you use it for all your work? And only a few hands raise, and I say, (in a friendly, humorous way of course) ‘you're all a bunch of liars, because most surveys put it well north of 80 percent, depending on how you ask, of students using it to some degree or another.’ And I think just being able to have an open conversation that says ‘look, I know you're using it, what I want to do is have a conversation of the pros and cons of it. You're here to learn and develop skills - how can you do that if you're dumping your ideas or your learning onto something that is quite frankly dumber than you, and not to your own real values or needs?’ While it's obviously not a silver bullet, and they're not going to stop cheating all of a sudden, they're going to be more conscious of the choices that they're making, and then they will talk to their friends, which creates a groundswell on this stuff. You've got to have champions, people making the argument who are willing to get into the arena and start trying to convince more folks, because that's the only way any of this changes: you just have to convince more folks.

*Cullen, C. S., & Murphy, G. (2025). Inconsistent access, uneven approach: Ethical implications and practical concerns of prioritizing legal interests over cultures of academic integrity. Journal of Academic Ethics, 23(3), 1415-1434. https://doi.org/10.1007/s10805-025-09606-2