AI in tax: navigating an evolving regulatory landscape

The cost of complying with Australia's tax system is substantial, and artificial intelligence (AI) has the potential to reduce it. Taxpayers and their advisers are already using AI to gather, organise and analyse information, and to prepare first drafts of documents.

The Australian Taxation Office (ATO) has made it clear that it does not object to the use of AI, but it expects taxpayers to remain accountable for the outcome. It is now developing public advice and guidance on taxpayer use of AI in tax affairs. This includes proposed updates to its rulings on 'reasonable care' (MT 2008/1) and 'reasonably arguable positions' (MT 2008/2), together with a new fact sheet.

For organisations, this is a timely prompt to understand where AI is already being used across their tax functions. We consider that Responsible AI principles, built around human oversight and accountability, provide a practical starting point.

Can AI reduce the cost of tax compliance?

The ATO itself acknowledges the time and effort involved in getting tax right. In an August 2025 letter to the Department of Finance, the Commissioner of Taxation recognised that taxpayers find the tax system and the requirements of tax law complex and time consuming. In September 2025, the Government also asked the Board of Taxation to review ways to reduce red tape in the tax system, complementing the ATO's work on a compliance cost framework. 

Technology has long been part of the answer, and AI may be the most significant advance yet. Used well, it can take on repetitive, information-heavy work such as extracting and reconciling data, summarising large volumes of documents and preparing initial drafts for review. That frees taxpayers and their advisers to spend more time where their skills and judgement matter most.

Trust in AI is the foundation 

PwC's submission to the Joint Select Committee on Artificial Intelligence (14 September 2026) describes AI as "a significant opportunity to lift Australia's productivity, strengthen economic resilience and create new sources of growth". The submission also identifies trust as "the foundation on which everything else rests: providing the governance, assurance and regulatory certainty needed to scale AI responsibly and safely".

The submission also notes that while around two-thirds of medium to large Australian businesses report some AI use, fewer than one in ten describe that adoption as significant or at scale. In this environment, the efficiencies AI offers are likely to be realised only where the right governance and assurance are in place. 

This is particularly so in tax. A tax outcome depends on accurate facts, sound technical analysis and professional judgement applied to a taxpayer's particular circumstances. Using AI doesn't remove any of those requirements. It does, however, raise a practical question for every organisation: how do you make sure humans remain at the helm?

The ATO's position is taking shape 

Although formal guidance is still being developed, the ATO's public statements already give a clear sense of direction. The common thread is that AI is welcome as a tool, but it doesn't change who is responsible for the result.

Speaking at the UNSW 16th ATAX International Conference on Tax Administration on 8 April 2025, Second Commissioner Jeremy Hirschhorn described AI in these terms: 

The ATO's message to the wider community ahead of tax time 2026 was equally direct. In a media release on 27 April 2026, ATO Assistant Commissioner Anita Challen said that "taxpayers remain accountable for ensuring the information they or their agents provide to the ATO is accurate - whether the advice came from a friend, online sources, or if AI tools were used in its preparation".

The ATO has taken a similar line with the tax profession. At the Private Groups Stewardship Group meeting on 9 September 2025, the ATO said it "does not have an issue with the use of AI but expected human review of material produced with the assistance of AI before it is finalised". 

The ATO isn't alone. The Tax Practitioners Board, the Federal Court of Australia and legal profession regulators have each issued their own guidance on the use of AI. While their focus differs, they share a consistent theme: AI can support professional work, but it doesn't replace the responsibility of the people using it.

What the ATO's guidance is expected to cover 

Item [4296] of the ATO's Advice under development - administration issues program, titled 'Taxpayer use of artificial intelligence in tax affairs', sets out three proposed products.

These concepts matter because they sit at the heart of the administrative penalty regime for false or misleading statements and for positions that are not reasonably arguable. How these concepts apply where AI has contributed to a tax position may therefore affect whether, and to what extent, penalties are imposed if that position turns out to be wrong.

While the timing of the final guidance is yet to be confirmed, it's clearly on the way. Organisations using AI in their tax affairs should expect closer attention to how that use is governed.

Responsible AI in a tax context

The concerns raised by AI aren't insurmountable, and they shouldn't stop you pursuing the efficiencies it offers. Using AI responsibly can help you unlock value while managing risk and building trust. 

One of PwC Australia's five Commitments to Change is to "improve discipline and rigour of decision making". Consistent with this, the Responsible AI principles set out in the PwC Network AI Policy are:

• complying with applicable laws and regulations

• promoting fairness and ethical use

• respecting privacy and protecting data 

• supporting security and safety

• encouraging transparency and explainability

• promoting reliability and quality

• maintaining oversight and accountability, and 

• fostering innovation with integrity.

Applied to tax, these principles encourage you to look beyond the output of an AI tool to the whole process around it: what information goes in, what the output is used for, how it's checked, who is responsible for the final decision and what evidence is kept to support the result.

Not every use of AI calls for the same level of control. Using AI to organise documents raises different considerations from using it to research a technical position or to prepare figures for a tax return. A responsible approach is proportionate to the task and to the consequences if the output is incomplete or wrong.

Above all, the principles put people at the centre. A clearly identified person should remain responsible for each tax decision, understand the role AI has played, and be able to challenge, explain and, where necessary, override the output. AI can support the work, but it shouldn't obscure who is accountable for it. 

Questions your tax function should be asking now

You don't need to wait for the ATO's guidance to be finalised before taking practical steps. Start by considering the following questions.

• Where is AI being used in our tax processes today, including features built into existing software and informal use of publicly available AI tools that may not be visible to the tax function?

• What role does AI play in each case, from administrative support through to contributing to a position reported to the ATO? 

• What human review takes place before an AI-assisted output is relied on, and is it appropriate to the significance of the outcome?

• Could we explain to the ATO, if asked, how an AI-assisted outcome was reached and who approved it?

• How are facts, calculations and legal sources verified?

• How is confidential and personal information protected when AI tools are used? 

• Do we know how our tax advisers and outsourced service providers use AI when working on our tax affairs, given that taxpayers remain accountable for the accuracy of information their agents provide?

The answers can help you identify where existing policies and controls remain fit for purpose, and where further work may be needed.

AI can make tax compliance faster and less costly, but the ATO has been clear that responsibility for tax outcomes stays with taxpayers and their advisers. Its forthcoming guidance is expected to explain how long-standing concepts such as reasonable care and reasonably arguable positions apply when AI is used, through updates to existing rulings rather than a new set of rules.

Organisations that map their current use of AI in tax now, and test whether their oversight and controls remain appropriate, are likely to be well placed when that guidance is released. Responsible AI principles, with humans firmly at the helm, provide a practical foundation for that work.

We're working with organisations to understand the opportunities, risks and practical questions AI raises for their tax functions, so you can adopt AI with confidence. We'd also welcome your perspectives on the issues that would benefit from further clarity as the ATO's approach develops.