The Assured Assessment Standard

Authors

Professor Luke Bereznicki, Pro Vice-Chancellor (Learning and Teaching Performance), University of Tasmania

Dr Alan Finkel AC, Founder and Executive Chairman, Proudly Human

Version
2.4
Date:
18 November 2026

Preamble

1)

The Assured Assessment Standard is a joint initiative of the University of Tasmania (as the Lead Design Partner) and Proudly Human (as the Independent Auditor). It addresses the problem that university students can use generative AI to partially or substantially complete their assignments and other assessment tasks. When AI substitutes for the student's own work, a pass no longer shows that the student has built the knowledge, skills and capabilities their award certified. This situation undermines learning, and the confidence of students, communities and employers in the value of university awards.

2)

In most jurisdictions, accredited universities are obliged by their regulators to ensure that all graduates have personally earned their award.

  1. In Australia, universities operate within the Higher Education Standards Framework, and in the United Kingdom, within the requirements of the Office for Students.
  2. Across the world, many degrees must also meet the standards set by professional accreditation bodies.
  3. These obligations have become increasingly difficult to fulfill with the shift to remote study and take-home assignments, where students can use generative AI to do their assignments in minutes at little or no cost. Students who use AI this way can pass without achieving the intended learning outcomes.
  4. Universities cannot prevent the use of AI for unsupervised assignments by relying on AI detection tools because of the risk of false accusations and an escalating race between the AI-powered generation tools and the AI-powered detection tools.
Regulatory bodies have made it clear that institutions must assure learning by design and not through detection.
3)

There are numerous reports of significant and growing rates of use of generative AI by students to help with their assessed work.

  1. Some of this use is legitimate, allowed by educators and acknowledged by students as part of how they build capability.
  2. The problem arises when AI is used as a substitute for what needs to be the student's own demonstration of learning. The extent of the problem varies by discipline and by university, but without deliberate assessment design no university is immune to the unauthorized use of AI by students.
  3. Detection tools are inappropriate in the light of the possibly high rates of generative AI use by students in preparation of their assignments. Instead, the focus must be on designing and maintaining the secured conditions that assure learning.
4)

A related problem is that a university that implements policies and procedures to ensure that every graduate achieves every learning outcome has no simple way to indicate to the local community and to global employers that the testamurs (degree certificates) they issue are faithful proof of the graduates' achievements.

  1. The AssuredAssessment™ certification mark displayed on testamurs and marketing materials solves this problem by providing independent, instant, confirmation to regulators, communities and domestic and international employers that every graduate has personally earned their award.

The Certification Claim

5)

The Assured Assessment Standard is expressed in a single certification claim:

Every graduate has personally demonstrated every Course Learning Outcome (CLO) under secured conditions.
  1. The evidence that substantiates this claim helps with related regulatory requirements such as course approval, accreditation, external referencing and regulatory assurance.
6)

Secured conditions assure identity (the person assessed is the enrolled student), authorship (the work and reasoning are theirs) and human observation (an assessor sees or hears it happen).

7)

In-person assessment is the most reliable way to meet secured conditions, with identity verified and the student directly supervised by an assessor or invigilator.

  1. Examples currently considered secure for on-campus study:

    1. Observed skill demonstrations,
    2. Oral assessment or viva voce,
    3. Invigilated examinations,
    4. Supervised practicals,
    5. Laboratory demonstrations,
    6. In-class presentations with live questioning,
    7. Observed competency-based assessments,
    8. Observed work-integrated learning, and
    9. Secured decision points in programmatic assessment models.
8)

Online assessment meets the definition of secured conditions when identity and authorship are evidenced, and human interaction and observation are present.

  1. Synchronous, dialogic methods are the most secure options for online students. For example, interactive orals and viva voces on submitted work let the assessor probe reasoning in real time, providing evidence of authorship and observation. Combined with identity verification methods, these forms of assessment meet the definition of secured conditions.
  2. Examples currently considered secure for online study*:

    1. Live questioning about the activity,
    2. Interactive oral,
    3. Viva voce on submitted work,
    4. Live demonstration or practical,
    5. In-person secured task at a university campus or other approved center such as a community library or town hall,
    6. Developmental submissions with interviews,
    7. Real-time, observed task completion with interaction,
    8. Observed practicums in workplaces, and
    9. Observed practice by a known observer.

    * Online study may include on-campus or in-person assessment requirements.

  3. Online exams supervised remotely, by live video or by proctoring software, cannot on their own reliably exclude concealed assistance. Where a university uses this method, it must show how authorship is assured (e.g. by pairing the exam with a synchronous oral on the same outcomes) to meet the definition of secured conditions. Where an exam is the appropriate form of assessment, holding it in person or at an approved assessment center is preferred.
9)

Reasonable adjustments to accommodate individual learning needs may be made so long as they do not alter secured conditions.

10)

University autonomy is deeply respected. It is up to universities to design their own courses and assessments, consistent with government regulations and professional accreditations. The AssuredAssessment™ audit evaluates internally designed programs against the Assured Assessment Standard.

11)

To earn the AssuredAssessment™ certification mark, the university accepts an external audit, performed by Proudly Human. The audit process verifies the university's methodology, records and appeals processes. Staff training against the Assured Assessment Standard is assessed, and courses and assessments are randomly sampled and evaluated.

The Certification Tests

12)

The certification claim is held to four tests:

  1. Coverage. For every course approved by the university's peak academic body (including nested and exit awards), staff provide a curriculum map that shows how every CLO is verified at least once, by a secured, must-pass task or secured decision point. A Course Coverage Matrix is derived from the curriculum mapping, showing the nature of the assessment tasks used to verify each CLO. If the mapping surfaces gaps, they must be remediated.
  2. Sufficiency. Every CLO is verified at the Advanced stage, being the exit level of the award. In Australia this is calibrated to the Australian Qualifications Framework (AQF), ranging from Level 5 for a Diploma to Level 10 for a Doctoral degree. Elsewhere this will be the equivalent national qualifications framework.
  3. Progression. Key Assessment Tasks are the culmination of a staged sequence of learning rather than a one-off event. Each course learning outcome is introduced and developed before it is verified at the exit level.
  4. No exceptions. Every Key Assessment Task must be passed. It is not sufficient to look at aggregate marks, allow substitution of tasks or units, or condone a fail. Reasonable adjustments can be made if they do not alter secured conditions, developmental staging or the must-pass status of the task.
13)

A Key Assessment Task is a secured task or secured decision point in which the student personally demonstrates the outcome at the exit level of the award. It is a must-pass hurdle, regardless of graded weight. A secured hurdle below the exit level is not a Key Assessment Task, although it still counts towards Coverage. In the case of programmatic assessment, evidence accumulates across many low-stakes assessments leading to a high-stakes progression decision, made at a secured decision point, that serves as the Key Assessment Task.

The Learning Context

14)

The Assured Assessment Standard works with the Two-Lane Approach pioneered at the University of Sydney.

  1. Lane 1 assessment is the opportunity to evaluate each student's knowledge, discipline-specific skills, and 21st century skills assessed under secured, supervised conditions. Students may use generative AI where its use is directly observed and forms part of what the task assesses.
  2. Lane 2 develops the capability that Lane 1 independently verifies, supporting open, developmental learning where AI use, with acknowledgement, is allowed and encouraged to help students build the AI capabilities that will be required in contemporary workplaces.
The dividing line between the two lanes is observation and authorship, not the presence of AI.
15)

The Assured Assessment Standard is designed to work alongside the institutional graduate capability frameworks (or equivalent) to keep human thinking and creativity at the forefront of learning within the disciplinary context. These human capabilities, such as critical and systems thinking, creativity, ethical judgment and capacity to direct AI (not defer to it) will only become more valuable. Alongside discipline knowledge and skills, capability frameworks should make these human capabilities an important part of learning and use the Assured Assessment Standard to verify their attainment.

16)

The Assured Assessment Standard defines what must be verified, and accommodates programmatic assessment and evolving practices. Secured verification of every CLO is designed to sit within a body of evidence of learning, gathered across both lanes, over time and in authentic contexts.