AI fraud on the rise at law faculty: almost a third of all fraud cases involve AI
At the UvA Faculty of Law, 44 cases of AI-related fraud were reported last year. This accounts for nearly a third of all 143 recorded fraud cases. It is a striking increase compared to previous years, when AI-related fraud was still almost nonexistent
The Faculty of Law at the University of Amsterdam saw 44 cases of AI-related fraud and plagiarism in the past academic year (2024-2025). This is according to figures from the Examination Board. Students commit AI fraud when they use unauthorized artificial intelligence, such as ChatGPT, to complete assignments, exams, or theses.
In total, the faculty handled 143 cases of fraud involving both bachelor’s and master’s students, of which almost a third were AI-related. By way of comparison, in 2023-2024 there were 18 cases of AI-related fraud out of a total of 136 cases recorded. In 2022-2023, not a single case of AI fraud was detected out of a total of 129 cases of fraud or plagiarism. It appears that AI fraud is on the rise at the law faculty.
In 2024-2023, 23 cases were actually found to be fraud and resulted in sanctions, compared to 15 the previous year. In the remaining cases, the committee was unable to prove fraud or could not determine it with sufficient certainty. According to the law faculty’s sanctions policy, the severity of the punishment depends on the ‘seriousness of the fraud’ and the ‘degree of culpability of the student’. However, the exact sanctions that can be imposed are not specified in detail.
Law students convicted for AI use
Earlier this summer, there was a case involving a UvA law student who appealed against a sanction for AI use, reported Folia. The student had used ChatGPT for a group assignment, submitting four non-existent sources. After his lecturer discovered this, he confessed that the chatbot had generated the references.
The examination board labeled this as “serious fraud.” The student was excluded from the course and from the exams for two other courses. The administrative court upheld the penalty on appeal: the use of AI that makes it impossible to assess the student’s skills is contrary to the examination regulations.
This is not the only recent legal case. A master’s student in constitutional and administrative law was also punished this summer: she was suspended from her studies for six months after using AI on a large scale for a writing assignment.
UvA guidelines
According to the UvA’s guidelines, commercial AI tools such as ChatGPT may never be used in an educational context because they do not comply with the UvA’s privacy rules. Since September 1, 2025, the UvA has therefore had its own AI tool: UvA AI Chat, specially developed for students and lecturers to use generative AI responsibly within the educational context and which therefore complies with the privacy rules.
With UvA AI Chat, artificial intelligence may only support students, but may never take over their work. Submitting work that has not been written by the student is not permitted, unless explicitly stated otherwise. After all, lecturers must always be able to assess the insight, knowledge, and skills of the student themselves, according to the guidelines.
Broader picture lacking
Folia also asked other faculties how many cases of fraud were AI-related. This did not yield a clear picture: several faculties indicated that they did not (yet) have an overview or did not respond. This leaves it unclear how big the problem is outside the law faculty. At the Faculty of Humanities, for example, there were 372 reports of fraud last academic year, with the ‘majority’ involving the use of AI, according to the examination board. However, it is acknowledged that this impression may also have arisen because there is currently a lot of talk about fraud involving AI within the examination board.
