Turnitin has published its first Learning Integrity Insights Report on the use of artificial intelligence in education, finding that institutions are moving away from AI detection alone and towards classroom integration and customisation.
Drawing on global qualitative and quantitative data from educators, students and institutions, the report points to a shift in how schools and universities want to manage AI. Rather than focusing only on whether students used AI, many educators now want clearer information about how and why it was used in a piece of work.
That reflects a broader move away from a simple yes-or-no approach to AI in the classroom. More than 60 per cent of recent customer feedback prioritised transparency in AI use, while institutions that had already begun integrating AI increasingly asked for ways to tailor approved uses to a specific class or assignment.
Annie Chechitelli, Chief Product Officer at Turnitin, said the company was seeing demand for more flexible approaches.
"There is no one-size-fits-all approach to what responsible use of AI in education looks like," said Annie Chechitelli, Chief Product Officer, Turnitin.
That comes as many institutions are still developing formal rules on AI. Fewer than half say they have an AI policy, despite evidence from customer discussions that clear and consistent rules help both educators and students.
Policy gap
The absence of formal policy appears to be a major barrier to wider, more consistent adoption. Where rules are in place, staff and students have a clearer basis for deciding when AI use is acceptable and how it should be disclosed. Where they are absent, uncertainty remains over AI's role in teaching, feedback and assessment.
The report also highlights a feedback gap. Teachers say they often lack the time to give students as much feedback as they would like, while students may not have enough opportunities to seek guidance during the writing process. AI could help narrow that gap, but hesitation remains around its use in feedback and grading.
Chechitelli said that hesitation depends heavily on context.
"Working with institutions on solutions that support their responsible AI use goals, we are gaining insights into what's working, and what's not. What we're hearing from educators is that those who have integrated AI into their teaching want more customisation and insight in how it can be used by their students. We are also hearing from both educators and students that their comfort with AI's presence in feedback and grading is situation-dependent," said Chechitelli.
Custom use
The emphasis on customisation marks a notable step in the debate over AI in education. Early concern centred on detection and misconduct, especially after generative AI tools became widely available. The new findings suggest a growing number of institutions have accepted that AI will be present in classrooms and are now focusing on guardrails, disclosure and assignment-level rules.
That does not mean older forms of academic misconduct have faded. Traditional plagiarism has remained steady even as attention has shifted to AI-generated content. Since April 2023, 6.16 per cent of submissions have contained more than 80 per cent similarity to existing sources. Of those highly similar papers, 91 per cent primarily matched other student papers, 7.8 per cent matched internet sources and about 1 per cent matched publications.
That pattern closely resembles the period before generative AI tools were released more widely. Between late 2019 and late 2022, 6.86 per cent of submissions contained more than 80 per cent similarity. Of those, 88.25 per cent primarily matched other student papers, 10.75 per cent matched internet sources and about 1 per cent matched publications.
Those figures suggest that while AI has changed the discussion around academic integrity, established plagiarism patterns have remained relatively stable. For institutions, that may reinforce the need to manage both issues at once rather than treating AI as a replacement for earlier concerns about copied work.
The report is based on customer discussions about how AI fits into learning, assessment and marking. Institutions are increasingly seeking practical ways to balance experimentation with controls, especially where teaching staff want room to set different expectations across subjects and assignments.
Turnitin works with more than 16,000 customers across 185 countries and territories. Fewer than half of institutions report having an AI policy.