How Artificial Intelligence (“AI”) is Affecting Tax Professionals

The Shift to Real-Time Compliance

In the global theater of tax administration, a silent arms race is underway. Tax authorities are no longer merely "collectors of forms"; they have evolved into "harvesters of data”. In the past, tax compliance was a retrospective exercise where tax professionals spent almost 80% of their time collecting and analysing historical data to prepare annual tax returns. Today, the "servant" has taken over these routine, repetitive, and mundane tasks.

Malaysia is currently undergoing what economists call "the late-mover advantage". By observing the struggles of early adopters, the Inland Revenue Board of Malaysia (“IRBM”) has bypassed incremental steps and jumped directly into a high-tech, real-time environment. The implementation of the E-Invoice regime, which began in phases in August 2024 represents a fundamental shift. Unlike traditional PDFs or paper invoices, the Malaysian model uses a Continuous Transaction Control (“CTC”) Model.

Under the CTC model, every transaction must be validated in real-time by the IRBM using XML or JSON formats. This allows the tax authority to have a "real-time perspective" into taxpayers’ business activities, virtually eliminating the lagging time between a transaction and its reporting.

In this case, the IRBM acts as a central clearinghouse. Every invoice is validated by the "servant" (‘the MyInvois Portal’) before it even reaches the customer. It is centralised, mandatory, and data-rich. One of the objectives for E-invoice is to eliminate the shadow economy by making data evasion mathematically impossible.

AI-Driven Audits: The Hunter’s New Tool

It is not just the taxpayer using AI, the "servant" is also working for the tax authorities. The IRBM and other global tax bodies are now using AI for predictive analytics and network research to detect fraud.

AI can identify fraud and suspicious trading patterns across thousands of entities in seconds. This is a task that would take a human auditor years.

AI can generate a risk profile of a taxpayer. The AI algorithms are able to determine which companies have a high probability of tax avoidance, shifting the audit focus from random checks to data-driven "targeted strikes".

Foreign newspapers and economic experts have predicted that white-collar jobs will be "swept away" faster than any previous economic transformation. However, a critical distinction is emerging where “AI replaces tasks, not necessarily jobs”. While the tax landscape in 2026 shifts under the weight of AI, this psychological foundation is fracturing. For veteran tax agents, the arrival of AI is not merely a technical update, it may be an existential challenge to their professional identity.

Automating the Routine

According to industry experts, repetitive tasks that are currently being replaced include but are not limited to the following:

1. Data entry and classification of expenses.
2. Drafting standard tax returns and basic reports.
3. Initial research on common queries.

The Irreplaceable "High-Level" Functions

What remains is the human’s judgement where high-level thinking that AI currently cannot replicate.

Tax consultancy often involves "reasoning backwards" from a goal to find a legal premise. This requires a deep understanding of the spirit of the law and the client’s unique business context.

Professionals are bound by a common set of values and ethics. A machine cannot feel "honesty" or "trustworthiness," nor can it be disciplined by a professional committee for misconduct.

As compliance becomes automated, the Malaysian tax professional must transition into a strategic advisor. They move from asking "What is the tax?" to "How should we structure the business to minimise risk or optimise the tax exposure?".

The Ethics of AI in Tax Planning

In the evolving architecture of global finance, the integration of AI into tax administration and planning is no longer a futuristic concept, but a present reality. Tax authorities, including the IRBM are increasingly deploying algorithmic systems to sift through massive datasets, identifying anomalies that would take a human auditor decades to uncover. However, as the "servant" of AI becomes more embedded in the "master" of tax law, a profound ethical and legal question arises: Can a machine ever truly understand human "intent"?

Tax law is rarely a matter of simple arithmetic. It is a discipline rooted in the nuances of purpose, commercial substance, and the "spirit" of the legislation. While AI excels at pattern recognition, it falters at the altar of subjective judgment. This analysis explores the ethical boundaries of AI in tax planning, focusing on the irreconcilable gap between data correlation and legal "intent," and why the need for the human element.

Section 114(1) of the Income Tax Act, 1967 specifically targets individuals who "wilfully and with intent to evade or assist any other person to evade tax" commit certain acts. The keywords here “wilfully” and “intent”, require a determination of mens rea (a guilty mind).

AI, by its very nature, is a probabilistic engine. It identifies that "Event A" (a discrepancy in reported income) often correlates with "Event B" (tax evasion). However, correlation is not causation, and more importantly, correlation is not intention. An AI might flag a taxpayer for a multi-million ringgit discrepancy, but it cannot determine if that discrepancy was the result of a complex corporate restructuring with a "reasonable excuse" or a deliberate attempt to defraud the state.

In addition, tax law is often about the interpretation of ambiguous terms. For instance, the determination of whether an expense was incurred "wholly and exclusively" in the production of income [i.e. Section 33(1) of the Income Tax Act, 1967] is a contextual exercise. An AI might see a high entertainment expense and flag it as a personal benefit. A human tax professional, however, understands the "intent" behind the expenditure. Perhaps, the high entertainment expenditure is a critical business-to-business negotiation that secured a project worth ten (10) times the expense. The machine sees the "what"; whereas the human understands the "why."

In a court appeal case, the human element is also paramount because, pursuant to Paragraph 13 of Schedule 5 of the Income Tax Act, 1967, the "onus of proving that an assessment is excessive or erroneous" lies on the appellant. AI cannot "prove" a case in a court of law, instead, it can only provide data. A tax agent must weave that data into a narrative that demonstrates the commercial substance of a transaction.

An AI might flag a Real Property Company transaction as a red flag for RPGT evasion. A human agent, however, can present evidence of a genuine corporate restructuring aimed at "IPO readiness" rather than tax avoidance. The agent uses Paragraph 14(b) of Schedule 5 of the Income Tax Act, 1967 (i.e. the appellant may be represented by an advocate or a tax agent or by both an advocate and a tax agent) to ensure that the "intent" of the restructuring is given weight over the machine's suspicion.

For the Malaysian tax professional, Paragraph 14(b) carries an implicit ethical burden. If a tax agent uses AI to prepare a client’s appeal but fails to verify the machine's output, they are in breach of their professional code of ethics.

Relying on a "bad answer" from a "good servant" (AI) does not absolve the professional of accountability. If an AI suggests a tax position that lacks "intent" or "substance," the agent who signs off on that position is the one who bears the legal consequences under Section 114(1A) of the Income Tax Act, 1967, which penalises those who assist or advise in the under-declaration of tax "without reasonable care".

AI is undoubtedly a "good servant." It can bring unprecedented efficiency to tax compliance and help authorities combat the shadow economy. But it remains a "bad master." It cannot grasp the concept of "intent", it cannot appreciate "commercial substance", and it cannot provide the "justice" that a human-led appeal process offers.
Paragraph 14(b) of Schedule 5 of the Income Tax Act, 1967 stands as a vital reminder that in the eyes of the law, the "human element" is not an inefficiency to be automated away. It is the very heart of the system. As we navigate the ethical landscape of AI in tax planning, our goal should not be to replace the tax agent with an algorithm, but to empower the agent to use the algorithm to tell a more accurate, more transparent, and more "intentional" human story.

The tax profession is not being destroyed; it is being redefined. The "manual, redundant processes" that once defined the tax professionals’ life are becoming extinct, replaced by more intellectually stimulating work.

We must accept technology as an enabler, not a disruptor. The successful tax professional of the future will be the one who uses AI as a tireless "servant" to handle the volume, while remaining the "master" of the final judgment. As the large firm’s incident mentioned earlier taught the world, the moment we relinquish mastery to the machine, we lose the very essence of what it means to be a "professional", which is the ability to provide an assurance of truth in an uncertain world.

For the Malaysian tax professional, the "Professional Mindset" must now encompass Digital Resilience. The outlook is positive for those who can transition from "Tax Preparation" to "Tax Technology Governance." The future veteran will not be the one who knows every section of the relevant Act by heart, but the one who knows how to ensure the AI is applying that Act correctly.

The Belanjawan MADANI 2026, tabled by Prime Minister YAB Dato’ Seri Anwar Ibrahim, reaffirms Malaysia’s commitment to building an AI Nation by year 2030. Malaysia will likely operate its own "Sovereign AI Cloud" for tax data, ensuring that the country’s financial intelligence remains under national control.

If we treat AI as a good servant, we free our minds from the shackles of manual data entry and mundane reconciliation. If we guard against it as a bad master, we preserve the human judgment, ethics, and accountability that no machine can replicate. The veteran tax agent of tomorrow is not a relic of the past, but the ultimate "Master of the Machine" that the only one with the wisdom to know when the AI is "hallucinating" and the courage to correct it.

As the saying goes, "Change is the only constant in life." For the tax professional, the key to mental health and career survival is to see AI not as a replacement for the self, but as a superpower for the soul.