Beyond Proctoring: How Cadmus is redefining secure assessment

For years, secure assessment has been treated as a surveillance problem. The assumption was straightforward: if students were watched closely enough—through live monitoring, screen recording, and rigid lockdown environments—assessment integrity would follow.

Generative AI has exposed the limits of this approach. Security cannot rely solely on monitoring student behaviour at a single point in time. Surveillance may deter some misconduct, but it does little to ensure an assessment is valid, fair, or reflective of how knowledge is applied beyond the classroom.

Why proctored exams are being replaced

As the limits of surveillance-based assessment become clearer, many institutions are questioning whether traditional proctoring truly upholds academic integrity. Increasingly, this approach feels out of step with how learning and assessment function today, and it brings with it a range of practical and ethical challenges:

• Limited efficacy: Research shows proctoring rarely improves academic integrity outcomes (Hussein et al., 2020; Zhao et al., 2024).
• High cost: Scaling traditional proctoring is prohibitively expensive.
• Privacy concerns & legal restrictions: Monitoring devices, webcams, and environments raises serious ethical and legal questions, with some jurisdictions prohibiting certain proctoring practices all together.
• Post-pandemic shifts: Many universities have moved away from proctoring entirely; for example, at the recent University of Melbourne x Cadmus Teaching & Learning Forum, most institutions reported that they no longer use proctoring.

Proctoring also creates a slippery slope: institutions often feel compelled to monitor devices constantly, block emerging cheating methods, and escalate invasiveness, creating an unsustainable cycle. Generative AI accelerates this breakdown, making attempts to block or detect its use increasingly futile.

The central question has shifted from “How do we stop students from using GenAI tools?” to “How do we assess thinking, judgment, and authorship in a world where AI exists?”

What secure assessment really means today

For years, “secure assessment” has been framed as controlling student behaviour, with proctoring seen as the main way to manage risk. Yet this approach cannot guarantee what matters most: that an assessment truly measures knowledge, skills, and judgment.

In a world where AI is everywhere and professional work is open-book, observation alone cannot ensure valid, fair, or defensible outcomes. At best, it shifts risk elsewhere, leaving fundamental questions about task design, authorship, and evidence of learning unresolved.

Today, secure assessment is about designing tasks that are robust by design—assessments where misuse is harder, learning is visible, and outcomes are defensible.

This means:

• Authentic tasks that reflect real-world thinking rather than artificial exam conditions
• Clear expectations around responsible AI use
• Assessment designs that prevent misconduct rather than react to it
• Opportunities for students to explain, justify, and reflect on their work
• Embedding fairness, accessibility, and trust into the assessment itself

How Cadmus replaces traditional proctoring with better design

Cadmus reduces reliance on traditional proctoring by designing out the problem.

Instead of concentrating all security at submission time, Cadmus supports assessments that emphasise process, reasoning, and decision-making across the learning journey. Students engage in staged writing, iterative development, and structured reflection, making their thinking visible over time. They are encouraged to articulate how ideas evolve, how sources (including AI) are used, and why particular choices are made.

By focusing on the learning process rather than just the final product, Cadmus helps educators design tasks that are harder to outsource, automate, or fabricate. In doing so, students’ authorship and judgement become visible, providing educators with meaningful evidence of learning.

Some controls remain, but only to support learning, not replace it. Cadmus is preventative and pedagogical, not punitive. Integrity is built into the assessment experience rather than imposed through monitoring.

When assessment is designed to surface thinking, explanation and judgement, one approach becomes especially powerful…

Making thinking visible: vivas and oral assessment

When assessment focuses on process, reasoning, and decision-making, it naturally raises the question of how students explain and defend their thinking. One of the most effective responses to generative AI is also one of the oldest: asking students to explain their thinking.

Vivas and oral assessment moments create space for students to articulate understanding, defend decisions, and demonstrate judgement. They make authorship explicit and significantly reduce the value of outsourcing or over-reliance on AI. Just as importantly, they reposition assessment as a dialogue rather than a transaction.

Why Cadmus is not a proctoring tool, what is does (and why it’s different)

Cadmus is not a proctoring tool because it does not rely on student surveillance to ensure academic integrity.

Instead, Cadmus supports learning-first assurance by helping educators understand how an assessment was constructed, not by watching students while they complete it.

Cadmus includes tools that support assessment integrity without surveillance (including our Learning Analytics):

• Locked browser options to reduce accidental or inappropriate navigation during assessments
• Analytics that flag unclear learning assurance behaviours, prompting educator review rather than automated accusations
• Assessment construction analytics, allowing educators to see how a student’s work developed over time

These features focus on evidence of thinking, authorship, and process, not monitoring students’ bodies, rooms, or devices.

What Cadmus does not do

Cadmus deliberately avoids invasive proctoring practices:

• No webcam monitoring
• No microphone recording
• No biometric tracking
• No automated “suspicious behaviour” judgements (educators retain full discretion)
• No screen recording or surveillance of student activity

This is intentional.

Learning-first security, not surveillance

We believe academic integrity is demonstrated through:

• Reasoning and revision
• Reflection and drafting
• The progression of ideas over time

Cadmus helps educators verify authenticity through learning evidence, not compliance under observation.

The future of secure assessment

Some institutions are already moving beyond proctoring to embrace learning-first approaches.

• Bond University treats AI as an opportunity to educate students on responsible use rather than automatically flagging it as misconduct.
• La Trobe University has adopted an AI-first approach, embedding AI across courses to support the learning journey and give students structured opportunities to develop reasoning, authorship, and reflection.

These examples demonstrate that the future of secure assessment isn’t about controlling students, it’s about designing assessments that make thinking visible, teach responsible AI use, and provide evidence of authentic learning.

The future of assessment is not more control, it is better design.

As AI reshapes education, institutions that succeed will move beyond proctoring, embrace authentic assessments, foster reflection and oral justification, and design learning-first, defensible, and secure assessments. Cadmus helps make that future possible.