Professor denounces mass AI fraud on an exam at Brown University: 'Academic integrity is at risk'

[highplainsdem]

https://english.elpais.com/education/2026-06-28/ai-fraud-at-brown-university-academic-integrity-is-at-risk.html

This year, the economist decided that both the midterm and the final exams for his course would be of the take-home, closed-book type (there is a certain tradition of this at Ivy League schools). “It’s a very nice kind of exam, because as you’re giving students practically unlimited time to complete it, it lets you make it harder than normal, to see how far they can go.” In this case, Serrano changed some of the model assumptions they had seen in class, and asked students to demonstrate whether certain statements were true or false under the new assumptions.

The course, which he has been teaching for years, is not an easy one: it typically attracts few students, but very good ones. He has never had more than 30 students enrolled at a time, and on some occasions he had only eight. This semester, probably because of the new evaluation system, 86 students signed up for the class. The results of the midterm exam, which was administered on March 5, were extraordinary, with an average score of 96 out of 100. Forty students scored a perfect 100. The people who corrected the exams warned him about several irregularities. “Some answers contained unusual passages that coincided with results obtained after running the questions through ChatGPT,” he says.

Serrano did not void the midterm exam, but warned students that the final one, which counted for 50% of the final grade, would be held in-person. He also said that if the grade distribution was not similar to the midterm, only the final exam would be taken into account. The average score dropped to 48 out of 100. Of the 89 students who did the midterm exam, only 59 showed up for the final one. And of the 27 who did not show up, 22 had scored a perfect 100 in the midterm exam.

“The empirical evidence of fraud is overwhelming,” says the professor, who has decided to make changes for the coming academic year. First, the weekly exercises will not count towards the final grade, as these could be done with AI. Second, no more take-home exams, no matter how appropriate they would be.

-snip-

This is damning for those students who used AI to cheat, all of whom showed a complete lack of integrity.

And it's damning for university officials, who initially didn't respond at all to his reporting the cheating. This article mentions that Serrano worries that Brown students "always get the benefit of the doubt" because so many of them have wealthy parents who donate to the university.

He doesn't mention - though some quick searching I did confirmed - that Brown's officials have involved the university in a number of AI partnerships and programs. For example: https://www.browndailyherald.com/article/2025/11/brown-begins-to-implement-university-wide-ai-tools . So those officials not wanting to respond to clear evidence of students cheating with AI probably has a lot to do with that entangelement with AI companies. Which makes me wonder how many school officials corrupted by AI companies are trying to suppress reports of cheating with AI.

Btw, the newspaper quoted here, El Pais, is a liberal paper, one of the largest newspapers in Spain, and one that does a lot of reporting on international news as well as Spanish news. https://en.wikipedia.org/wiki/El_Pa%C3%ADs

[Attilatheblond]

1. Most enlightening information, thank you for bringing it here for us

Sadly, not really a surprise that entitled students who worked the system get away with it due to financial status along with college's partnership with AI programs.

The saddest part is this is an economic class that generally provides the most 'leadership' class in our nation and that means cheaters who are basically not really educated rising to powerful positions. Not really helpful to our national interests to have lazy entitled brats growing up to be in charge.

[ImNotGod]

2. Who needs higher education when you can fake your way through life. nt

[2naSalit]

9. As expressed by a member of...

The nonpresident's legal team...

When asked whether she'd rather be smart or pretty she answered, "Pretty, I can fake 'smart'".

[SergeStorms]

23. To someone....

of equal or lesser intelligence you might be able to, "fake smart."

In the Trump administration it's a given.

[dalton99a]

3. There is a reason med schools don't allow take-home exams

You don't want a doctor who graduated by cheating and looking up stuff

[RandomNumbers]

4. I want to know which AI tool they were using

and how they got such high accuracy, when my luck with using AI on my job (perfectly legit, although I have to be careful to avoid anything proprietary in my prompt), is not quite that good. I often get flat out bullshit stuff in the responses. Although overall I usually get some benefit for the time expended, even if it is just new ideas to explore.

Basically I think I only have access to Copilot on my work pc. Maybe it's time I figured out how to use ChatGPT or Claude. (I'm cheap though and refuse to spend any money)

[highplainsdem]

10. Those AI models you mentioned were all illegally trained on stolen intellectual property. I'm sorry

you're forced to use AI for your job, but they are all unethical tools, and the AI companies should be sued out of existence, and those responsible for the IP theft should be prosecuted and should get prison sentences.

As for how those disgusting, cheating students got some correct answers - there was apparently no time limit or supervision for those cheaters, and they had lots of time to cross-check results, and could have used any number of different AI models, multiple times.

And all those AI-using cheats are dumber as a result, and more dependent on AI.

[RandomNumbers]

13. Not necessarily in the case of how I use it for work.

Although to be clear, I 100% agree with your disgust at the unethical uses of AI.

So far I have only ever seen public forum posts used in responses by Copilot to my prompt. It turns out it is faster to use Copilot than to search those forums myself. (I only use it for VERY specific types of queries related to a very popular software platform that has a massive amount of public web content.)

I have to take what Copilot gives me and make it actually WORK in manually written software solutions (we can't use AI agents yet - GOOD!). Half the time it is no better than a pointer to other search terms for me to follow up with more detailed manual searching.

My biggest concerns with AI are:
* environmental impact of data centers - this COULD be managed if we have the will, which sadly we won't in most places
* use of AI for unethical goals - I like Anthropic's focus on guardrails and wonder if it's enough, but Trump hates them, so they must be better than the alternatives
* your last point - that use of AI can make people dumber and more dependent on it than their own brains. I think this could also be managed, but first people have to recognize the potential issue.

[ancianita]

31. Cory Doctorow teaches us what AI can and cannot use, and why most of what AI uses is not stealing intellectual property.

I highly recommend you read his latest book.

[highplainsdem]

34. I like Cory and often agree with him, and he'd sometimes repost my comments when I was still

posting on X. But I disagree with him on this. AI is trained on stolen intellectual property, and people are dumbed down by using it, no matter what excuses they make for it - and it's exremely harmful tech, the most harmful non-weapon tech ever developed.

[highplainsdem]

35. And btw, Cory isn't in agreement with most creators of intellectual property on copyright:

https://en.wikipedia.org/wiki/Cory_Doctorow

He is an activist in favour of liberalizing copyright laws and a proponent of the Creative Commons organization, using some of its licences for his books.

[mopinko]

12. probably cuz it's a narrow subject.

not a lot of contrary writing out there about it.
on any more popular topic, there’s a lot of dreck.

[cab67]

22. Narrow subjects aren't necessarily immune to the problem.

I've done some experimentation in my own field (which is also very narrow), and the results I obtain are hilarious.

[LAS14]

5. Clever prof! nt

[msongs]

6. universities will take the money, thus prostituting their values by bending over for trump

among other things

[cab67]

25. It's not necessarily a matter of politics.

Or if it is, it doesn't necessarily have to do with the president.

Many universities are being approached by tech companies to use their AI services as an adjunct to other modes of education. These can be expensive, so universities are offered discounted "educational" rates, and administrators are being told it's the way of the future. There is thus a disconnect between some (not all) in the administration and most (though sadly not all) faculty when it comes to this sort of thing. Faculty are being encouraged to use AI, or to let students use it, in spite of the clear evidence that (a) AI is unreliable and (b) even when it's reliable, it's harmful to the educational process.

A survey was sent to us a couple of weeks ago asking how we integrate AI in our classes. My answer was very brief - I forbid its use.

The same thing happened when instructional technology first became a thing in the 1990's. I was happy to use PowerPoint for pictures because it meant I wouldn't have to sort through slides and put them away, but otherwise, I saw little use in some of the multimedia packages we were encouraged to use. All I could see was that students stopped taking notes the minute some sort of digital resource was made available. DIdn't have to be online, either - it could have been a CD-ROM that came with a textbook.

I haven't really been hit with pressure to use AI at my own institution, but I have colleagues elsewhere who have. It's about money, yes, but not necessarily about national politics.

[Coventina]

7. And yet, certain DUers think this is just fine and we're yelling at clouds.

[Disaffected]

15. Certain DUers think using AI to cheat on exams if "just fine".

I've yet to run across such posts. I'd be interested is seeing a few examples.

[Coventina]

16. thread from a few days ago

https://www.democraticunderground.com/100221330017

[Disaffected]

18. OK, I skimmed that thread and didn't see any posts

that condoned cheating by AI (or any other way). Care to quote the offending ones?

[Coventina]

19. I'm not going to call out specific people, as that is against DU TOS.

But seriously, I posted a thread about academic integrity, and was met with posts dismissing my concerns and scorn about the use of blue books.

If you cannot draw inferences from that, I cannot help you.

[Disaffected]

20. Sorry but I cannot draw inferences that are not there.

And your accusation was not inferential at all.

I'm not asking you to call out anyone - simply state in your own words how someone(s) in that thread (or any other DU thread) said or implied that cheating on exams, using AI (or not) is "OK".

[Coventina]

28. Are you seriously making a case that coming into a thread about AI threats to

academic integrity and critical thinking and making remarks like

"precise skills are less relevant" regarding the use of blue books,

and comparing the use of calculators to the ability to write papers is NOT dismissing cheating / the ability to perform in an academic setting?

If that is truly your case, we have nothing to discuss.

[Disaffected]

29. That's some serious goalpost moving there.

I guess we do have nothing to discuss.

[yardwork]

8. This is sad and worrying.

The professor came up with a good solution, though.

Maybe universities need to bring back blue book exams, where students hand write the answers in a small book that is handed out in class.

[mopinko]

11. it's a brilliant solution imho.

i’ve read that some unis r going back to oral exams.

[Kaleva]

14. Yes

[sop]

33. I remember the dreaded individual oral exams administered by no-nonsense Jesuits at the end of each semester,

blue books were used for testing in most subjects, and no calculators were allowed in math courses. God, I'm so old!

[Mossfern]

38. But they can't write cursive!

Exams would take forever. to answer in print.
Maybe supply keyboards and word processors with no internet capabilities would work though.

[Disaffected]

17. Take home, closed book exams??

Seems an invitation to cheating, AI or not.

How do you prevent collaboration or paying someone to take the exam for you, let alone not opening any books?

Depending on the honour system to prevent cheating (if that is the underlying assumption) seems a bit quaint now-a-days or any-other-days for that matter.

[cab67]

21. It's right on my syllabus.

"The use of AI in the completion of any homework assignment is not allowed."

I'm sure students are using AI anyway.

Sometimes, the joke is on them. In my upper-level class, I can identify users of AI because they all have the same borderline-funny absurdly wrong answers. All ethical concerns aside, AI is not reliable.

[Disaffected]

27. This gets a bit murky but,

Last edited Sun Jun 28, 2026, 05:58 PM - Edit history (1)

how do you feel about student use of search engines to find web information relevant to home work or exam questions?

[cab67]

30. wary.

For the subjects I teach (vertebrate paleontology, with an emphasis on dinosaurs in one course), there are all kinds of online resources. Some of them are reliable, but many are not. They're put together by dinosaur groupies (they're a thing) who've memorized lots of details, but don't really understand the subject well enough to understand (a) what these details really mean (beyond "would dinosaur A kill dinosaur B, even though they lived on different continents and one was extinct 50 million years before the other appeared" ) and (b) how much nuance these details require given what's actually in the professional literature.

About 10 years ago, I published a paper naming a new species of extinct crocodile. It got a little bit of media attention. I checked Wikipedia from the moment the paper was released, and someone had posted a page about the new species one hour later. I had nothing to do with it, nor did any of the coauthors - and it was full of errors.

Same thing happened earlier this year. We published a new species of fossil crocodile, and there were a few news reports about it. It took slightly longer for the Wikipedia page on this new croc to come out, but it was so full of errors and misinterpretations that I almost asked its author to just pull it down and start over. Whoever wrote it made all kinds of non-sequitur statements and clearly didn't know (a) what we'd done or (b) what we'd concluded.

The great thing about the Internet is that anyone can post things. The problem is that anyone can post things.

Upper-level undergrads and grad students can usually discern the reliable from junk, but for beginning students or non-majors, it's not so easy.

[Disaffected]

36. Interesting and well put.

Thanks.

[SergeStorms]

24. If only....

this was going to be the worst consequence of using AI.

I'm not condoning the use of AI in any educational situation, but there will be far worse consequences from its use.

[erronis]

26. Just based on my reading history, El Pais is another fantastic non-US M$M outlet

Its Engilsh pages are well written and factual. I've added it to my normal scans of DW.com (Deutsche Welle), The Guardian, and others.

[Matthew28]

32. The only way to deal with this

is to allow companies to fire frauds. If you don't have the skills and you're proven a fraud. Well, you're fired. This should go on your record for life.

Any degree should also be considered a fraud.

[kentuck]

37. I find it fascinating...

...that the younger generations seem to look at AI differently than the older generations, it seems to me. They see AI as nothing more than a tool to make their jobs easier. They see no danger in it.

It seems to be the older generations that are skeptical of the AI models and the effects it may have on all of us.

Just my opinion.

[Mossfern]

39. Older but wiser

[anciano]

40. I think that the younger generations

realize that AI proficiency is already becoming a prerequisite for the jobs and careers of their futures.

[cab67]

41. This is a problem.

[highplainsdem]

43. See reply 42.

[highplainsdem]

42. I guess you've missed the news stories about polls showing younger users have a lower opinion of AI

than older users have.

For example:

Gen Z Is Using A.I., but Doesn't Feel Great About It.(NYT, 4/9)
https://www.democraticunderground.com/100221159855

Americans Have Turned Against AI in Incredible Numbers (Futurism, 6/21/26)
https://www.democraticunderground.com/100221320774


Those students at Brown cheating with AI knew damn well it was cheating. They thought they could get away with it anyway.

[anciano]

44. I think the appropriate and responsible use of AI

should be taught in the classroom. AI is not only changing the way we acquire, process, and manage information, but it increases our knowledge and enhances our creativity as well.

[highplainsdem]

45. Generative AI tools are unethical because they're illegally trained on stolen intellectual property.

They dumb users down, as studies have shown. They don't increase knowledge or enhance creativity - they offer a simulation of both. A pretense. A pretense filled with errors - hallucinations - unless the AI results are checked carefully and corrected.

One way teachers have been able to make it obvious that students are cheating with AI is to question them on the subject they supposedly learned about. The cheaters using AI are still ignorant, and can't remember much of what the AI generated for them.