The conclusion was that the sector keeps having this conversation about supporting international students, and there are pockets of excellence, but there’s little evidence of consistent improvements. And that’s a conference wrap. Let’s hope we continue to see support changes before #ICAI2027.
There’s already a lot of research about supporting international students at university level. It just isn't used enough. Little research exists about supporting international students at high school level, or how to maximise the existing knowledge of students when they transition to university.
Staff training and engagement remains a massive challenge. If we can't get staff to engage properly with integrity processes, we certainly can't expect students to. In the US, it was said to be difficult to have this dialogue within universities at present due to funding and structural challenges.
Students won't read written information but will watch someone reading out that same written information in a video, even if it takes longer. The format matters more than content. Everyone multitasks now (even conference delegates) and training materials must account for this.
Another shared example. International students were reported to have plagiarised their answers for a plagiarism induction course. They didn't see the topic as important. Students said to be particularly lacking in their ability to paraphrase without AI.
Students need to be supported to learn how to learn. We assume they arrive with study skills that transfer across languages and cultures. Often this isn’t the case. Doctoral students said to find the transition an under-appreciated challenge, particularly if they haven’t studied in English before.
A related point discussed if staff can't demonstrate that international students have been trained on integrity expectations, should the penalties they’re awarded be lower? It can be hard to justify full sanctions if a university hasn't done its part.
Should international students receive lesser penalties given they often have less support in a language that’s not native to them? This led to debate. Disruption to study plans by international students has a disproportionate impact.
A real case was discussed where a student with a proven misconduct finding didn't understand the outcome email message. If an integrity process can't communicate plain English outcomes to the students it is judging, that’s a fundamental procedural flaw.
Staff bias came up during the session. International students are sometimes reported for misconduct simply because they performed better than expected. That challenge runs deeper than just being an #academicintegrity concern.
A frustrated student holds a failing grade amidst symbols of academic integrity, legal issues, and global education challenges.
My final session at #ICAI2026 was another breakout group focused on working with international students, an overrepresented group in academic misconduct cases. There was a lot of frustration expressed in the room, but not many solutions.
This was a session based more around challenges than solutions, and ethical challenges around access to AI were raised. There are real benefits to AI for students with disabilities that we shouldn't be blocking. Many in the room supported the view that AI access should be enabled by default.
Handwriting was discussed for its knowledge retention benefits, but not everyone has been taught cursive. One delegate admitted having to look up letter formations. We can't assume handwriting is a universal fallback, and this requirement may also disadvantage some students.
The school-to-university transition is a mess. Students feel there’s clearer AI guidance at high school level than at university. Staff running welcome sessions can't answer basic questions about AI use, or say why working practices students rely on from high school are no longer allowed.
The Grammarly problem came up. Where does writing support end and AI writing begin? Very few universities have properly had this conversation. Students are using tools like Grammarly that their universities subscribe to and encourage the use of, but are then being called into disciplinary hearings.
Universities are supplying AI tools to students without updating academic integrity processes, training staff, or training students. Students are often way ahead of staff in understanding what AI can do. There’s very little awareness of AI agents (which I ran a session).
Detection bias against some student groups is still a live concern. Providers are increasingly training with corpora of international student writing to address this, but neurodiverse students are much harder to identify and account for. The bias problem isn't close to being solved.
Staff are also said to be bad at documenting evidence of AI misuse, sending investigators cases with unclear evidence trails. Programming is the biggest contributor of AI misuse cases at some universities, and the staff investigating these cases can’t fairly address this. They aren't programmers
AI detection and when to use it remains the number one challenge universities want solved. The desire for a magic tool hasn't disappeared, even as the evidence mounts that we're not there yet. One online university banned detection entirely, but is still seeing nearly 2,000 AI misuse cases monthly.
A vibrant graphic features the phrase "All Things AI," emphasizing themes of inquiry, learning, and brainstorming in technology discussions.
Spending my Sunday morning concluding #ICAI2026. Joined the All Things AI breakout group and the discussion covered a lot of ground, but it was clear to me that many delegates just hadn’t been able to keep their #GenAI understanding up to date. Here are some highlights.
Grace and Akshay talk about their honour council work and how they deal with academic integrity breaches (very different to the UK system). I like that community service is a standard penalty, 10 hours for a first offence, 30 hours in some cases.
Presentation slide titled "Building Academic Integrity Across Campus" with details about presenters and the session's date and time.
A presenter stands at a podium with a laptop, while a colleague observes, both engaged in a session against a blue backdrop.
And, now back to largely human written posts. Enjoying my final #ICAI2026 parallel session presentation from two student presenters, Akshay Valoor and Grace Feng.
My final advice for educators from #ICAI2026 - get familiar with AI agents and try them out. Students will increasingly hand in work created through multiple agent interactions, and we cannot see all the reasoning behind the final output. Assessment needs to adapt. This is where we're heading.
Abstract shapes represent a diverse audience engaged in discussion, emphasizing themes of connection and communication.
Closing my #ICAI2026 talk with a thought, asking will AI eventually understand more than humans about academic integrity? There’s an irony of building systems that might surpass us. Very grateful to have had a human audience. I’ve spend a lot of time recently talking to AI agents.
Mentioning Kimi Agent Swarm at #ICAI2026. This is one of the more capable agent systems available now. I used it to methodologically identify the top 100 academic integrity books. Some of the people in the room were on the list.
Abstract design featuring flowing waves in blue and orange tones, intertwined with geometric shapes and lines on a light background.
Discussing AI agents at #ICAI2026 and the pace of change is genuinely hard to keep up with. What I could demonstrate six months ago barely scratches the surface of what's possible now. Four examples made it into today's talk from around a hundred experiments. There’s a lot of research out there.
Abstract design featuring geometric shapes, graphs, and documents, representing data visualization and academic themes.
Sharing some data at #ICAI2026. 31% of 2025 academic integrity papers had AI terms in the title, and 26% focused on education. I used ChatGPT to turn the research into slides for my academic integrity research class. The irony of using AI to present on AI isn't lost on me.
Abstract design featuring interconnected screens, charts, and gears, symbolizing the integration of AI in education and digital processes.
Mocked-up newspaper headline discusses an AI named Einstein logging into a student system, highlighting rapid advancements in AI technology.
At #ICAI2026 discussing how AI agents have gone mainstream faster than expected. Referencing a mocked-up newspaper (created by an agent) about Einstein AI logging into Canvas as a student, watching videos, doing the work. The hypothetical is becoming reality quicker than we anticipated.
At #ICAI2026 describing my live transcription workflow, dictate into my phone, send this to OpenClaw to draft tweets with images, then queue this for review. I'm not auto-publishing because I want to check what Zak produces first. It's not working perfectly. That's rather the point.
A presenter stands at a podium, addressing an audience while a visual display shows data and graphics related to AI in education.
A hand holds a pen over handwritten notes while a robotic hand writes on a tablet, illustrating AI and human authorship in research.
Just presented at #ICAI2026 on how AI agents might actually function in education, including practical applications and genuine limitations. I also had an AI agent generating social media posts live while I presented. A few examples will follow.
