# Study Finds AI Models Outperforming Law Students in Recent University Exams

> Researchers at the University of Wollongong found that generative AI models have significantly improved in legal reasoning, raising concerns about academic integrity in unsupervised assessments.

- Published: 2026-09-30T14:00:01.000Z
- Category: science
- Source: [Phys.org](https://phys.org/news/2026-09-ai-exams-universities.html)
- Canonical URL: https://manypress.com/article/study-finds-ai-models-outperforming-law-students-in-recent-university-exams
- Publisher: ManyPress (https://manypress.com). When citing, link the canonical URL.

---

A new study from the University of Wollongong tested nine AI models across two compulsory law subjects, finding a significant improvement in performance compared to 2023. The AI tools, which had internet access but no subject-specific materials, outperformed the majority of students in both criminal law and tort law exams. Researchers noted that while AI capabilities have advanced, the models still exhibit inconsistent performance and occasional errors, prompting calls for new university assessment strategies.

### Key numbers

- **76.3%** — Average score for AI in criminal law
- **66%** — Average score for AI in tort law
- **82.5%** — Percentage of students outperformed by AI in criminal law
- **61%** — Percentage of students outperformed by AI in tort law

**Why it matters:** The study indicates that generative AI has reached a level of sophistication where it can outperform most students in complex legal exams, threatening the integrity of unsupervised academic assessments. This forces universities to rethink how they evaluate student knowledge, shifting toward models that require students to demonstrate independent reasoning and the ability to verify AI-generated content.

## Significant Gains in Legal Reasoning

In the 2023 criminal law exam, AI models averaged 52.5% and performed at the 22nd percentile. In the new study, AI papers in criminal law averaged 76.3%, outperforming 82.5% of students. For tort law, the models averaged 66%, outperforming 61% of students. Seven of the 18 AI-generated papers ranked at or above the 90th percentile of student results.

## Inconsistencies and Hallucinations

Despite the improved grades, the study found that AI models remain unreliable legal experts. Performance varied between subjects, with some models showing strong analysis alongside poor citations or fabricated authorities. Researchers observed that while some models had lower rates of hallucination, they also cited fewer sources overall.

## Proposed Assessment Changes

The researchers suggest that universities move away from reliance on unsupervised assessments. They propose a three-pronged approach: maintaining AI-free, monitored in-person exams; requiring students to collaborate with AI while grading the process rather than just the final product; and using a 'relay' model where students alternate between AI-assisted drafting and independent, supervised critique of that work.

### Timeline

- **End of 2022** — ChatGPT was released, prompting university concerns regarding AI-assisted cheating.
- **2023** — Researchers conducted an initial experiment testing AI on a criminal law exam.
- **2026** — The study 'Legal Minds Vs. Neural Networks' was published in Law, Technology and Humans.

**Key points**

- Researchers tested 18 AI-generated papers across criminal law and tort law subjects at the University of Wollongong.
- AI models outperformed 82.5% of students in criminal law and 61% of students in tort law.
- Seven of the 18 AI papers ranked at or above the 90th percentile of student performance.
- The study identified 'jagged capabilities,' noting that models could be excellent in one subject but weak in another.
- Researchers recommend a mix of supervised, AI-free exams and assessments that evaluate a student's ability to critically supervise AI output.