Artificial intelligence (AI) is routine in accounting education, it helps in accelerating feedback and access to

four propositions tied to key risks: over-reliance, privacy and security, bias and fairness, and transparency and

accountability. The output is an integrated framework combining a TPB model with an Ethical-AI issues map.

Implications include practical programme guidelines, MIA–university professional development and AI-aware

assessment to raise PBC, that convert intention into practice.

Keywords: Academic integrity; Accounting education; Ethical AI; Professional judgment; Theory of Planned

Behaviour (TPB)

The educational sector is among the most prominent that has received the benefits of advances in Artificial

Intelligence (AI). Academic assistance can be made available to students continuously through the use of AI

chances (ACCA, 2023). Recent discussions have focused on the need for investigation and supervision in order

to protect academic integrity and uphold ethics against the possible transformative power of (AI Qadhi, Al-

Duais, Chaaban & Khraisheh, 2024).

In addition, Al-Zahrani and Alasmari (2024) state that a strict, well-defined ethical rules are needed to deal

with privacy, security, and bias when utilising AI in higher education. Chechitelli (2023) reported that, among

38.5 million submissions analysed for AI-generated content, approximately 9.6% contained over 20% of text

likely produced by AI writing tools, while 3.5% exhibited between 80% and 100% AI-generated content.

In the context of accounting education, where ethical behaviour and professional judgment are important, an

overreliance on AI may damage students' critical thinking ability and professional judgment. The use of an AI-

driven decision-making system proved to improve efficiency according to Lehner, Ittonen, Silvola, Ström &

education.

Across accounting classrooms, students use AI for information seeking, summarisation, feedback, and solution

checking. Benefits co-exist with risks that are acute for the discipline: (i) academic-integrity breaches

(undisclosed assistance, ghost-writing), (ii) privacy and data-security exposure, (iii) opaque model bias that

can mislead analysis, and (iv) over-reliance that weakens professional scepticism. These issues warrant

curricular and assessment-design responses where it beyond discipline alone, so that responsible AI use

supports, rather than substitutes for, learning and judgment. Therefore, the current study frames ethical AI use

through TPB to guide targeted curricular and assessment interventions.

in accounting education with the aims to:

attitudes toward ethical AI use in accounting education.

policies) influence intentions to use AI ethically.

design affect students’ perceived control over using AI ethically and their professional judgment.

This study follows a structured conceptual analysis to build a TPB-grounded framework for ethical AI use in

Inclusion criteria focused on higher education contexts that discuss AI ethics or integrity, accounting or closely

related disciplines, studies that speak to attitudes, norms, or perceived control, and policy or competency

statements relevant to Malaysian programmes. Titles and abstracts were screened, followed by full-text review.

Evidence was then organised using a framework synthesis: each source was coded against TPB constructs

(attitudes, subjective norms, perceived behavioural control), intention, behaviour, and professional judgment,

and mapped to four discipline-salient issues, namely over-reliance, privacy and security, bias and fairness, and

transparency and accountability. Iterative team discussion resolved coding disagreements and refined the

proposition wording until consensus.

The synthesis produced two artefacts: an Ethical-AI issues map that links the four risks to TPB constructs, and

a TPB-based conceptual pathway from A, SN, and PBC to intention, behaviour, and professional judgment,

with associated propositions for future empirical testing. The approach is conceptual rather than exhaustive; it

privileges recent scholarship, English-language materials, and publicly available Malaysian policy documents,

consideration process that is influenced by three main determinants, which are attitudes, subjective norms, and

perceived behavioural control. In the context of using technology adoption in education, TPB is used

frequently to understand the acceptance and use of technology by students and educators.

In the context of Malaysia’s higher education system, the use of technology such as AI, Blended Learning

(BL), and mobile learning can be explained using the TPB. Students who possess a positive attitude towards

technology (considering AI as a tool to enhance learning experience) tend to have the intention to use such

technology. Whereas, the subjective norms, such as the national policy of Malaysia Education Blueprint for

Higher Education (MEBHE) 2015-2025, that encourage the use of ICT and support from peers and lecturers,

put social pressure that strengthens that intention. Meanwhile, the perceived behavioural control was

influenced by students’ digital literacy, ease of access, and perceived efficacy towards the use of technology, as

introduces four discipline-salient risks known as over-reliance, privacy & security, bias & fairness, and

transparency & accountability. All the risk that can be interpreted with the Theory of Planned Behaviour

(TPB). In TPB terms, students’ attitudes (A), subjective norms (SN), and perceived behavioural control (PBC)

shape intentions (I) and behaviours (B) around ethical AI use.

AI can improve efficiency and access to real-time help, but excessive reliance can blunt analytical and

problem-solving skills essential to accounting (Albrecht & Sack, 2000). TPB linkage: risk-heavy attitudes

reduce intention to use AI ethically (P1), while clear rules and assessment designs that require reasoning (orals,

workings, verification logs) strengthen PBC and ethical behaviour (P3).

Models trained on biased data can yield unfair outcomes in learning and assessment (Mohammed & Malhotra,

2025). Predictive systems may also threaten autonomy and due process (Akgun & Greenhow, 2022; Regan &

Jesse, 2019; Citron & Pasquale, 2014). TPB linkage: recognising bias elevates risk-salient attitudes (lowering

intention unless safeguards exist) (P1); strong institutional norms (fairness policies, audit trails) raise intention

(P2).

Educators report uncertainty over who is responsible for AI-driven errors for example developer, institution, or

lecturer in order to compromising accountability (Zawacki-Richter, Marin, Bond & Gouverneur, 2019). TPB

linkage: explainability requirements, disclosure templates, and decision logs increase PBC and strengthen

(SN) and perceived behavioural control (PBC) shape intention (I), how intention drives behaviour (B) in

coursework (e.g., disclosure, verification, limited reliance), and how repeated behaviours contribute to

professional judgment (PJ). Figure 2 presents the TPB-based model: A, SN, PBC → I → B → PJ, with a direct

perceived behavioural control (PBC) to intention (I), behaviour (B) and ultimately professional judgment (PJ),

the framework shifts the debate from “AI: yes or no?” to the more useful question of which institutional and

pedagogical choices move A, SN and PBC in real classrooms. Mapping the four discipline-salient risks, which

literacy, clear guidance and training, and simple procedures for disclosure and verification, should strengthen

PBC → Intention and help PBC → Behaviour in coursework. Second, clarity and consistency of programme

rules, which are what is allowed, when to disclose, how to verify, and how it is graded, should reinforce

Subjective Norms and help intentions translate into behaviour across assessments. Within Attitudes, students’

evaluation of ethical risk (integrity, privacy, fairness) is central to the A → Intention path; course activities that

surface and discuss these risks can shift that evaluation without expanding the model. These points refine how

the existing constructs operate in Malaysian accounting classrooms and provide straightforward targets for

peer expectations (e.g., disclosure rules, classroom culture), and perceived behavioural control captures

programme-level ethical AI guidelines aligned with MIA competencies and MoHE/Halatuju 4 strategies

(privacy, fairness, disclosure, accountability); (ii) build educator capability via MIA–university professional

development on responsible AI use and AI-aware assessment; and (iii) embed AI ethics across the curriculum

15. Mohammad, N. H., Zamri, N. A. K., Roni, M., Hadi, S. N. I. A., Sadikan, S. F. N., & Mahzan, S. (2025).

22. Zawacki-Richter, O., Marín, V. I., Bond, M., & Gouverneur, F. (2019). Systematic review of research on