It is becoming increasingly difficult to keep pace with the rapid changes AI is bringing to the tax world. Just this week, KPMG announced a partnership with Anthropic to embed Claude into its tax and advisory platforms, the first time one of the Big Four firms will integrate a generative AI tool so deeply into its core systems. Only a week earlier, HMRC announced a deal with Quantexa to deploy AI technology to improve customer service while also enhancing its ability to detect errors and fraud more efficiently. In fact, tax authorities are rapidly becoming some of the largest users of AI globally. So what does this mean for taxpayers? One thing is clear: both individuals and businesses will need to consider the significantly enhanced access to, and analysis of, data that tax authorities will now have at their disposal. Tasks that previously required teams of people manually reviewing tax returns to identify anomalies can now be performed in a fraction of the time across vastly larger populations of taxpayers. Inevitably, this is likely to increase the number and sophistication of tax audits. Greater transparency and data analysis capabilities will also help tax authorities tackle the “tax gap”, the difference between expected tax revenue and actual receipts.
AI will allow tax authorities to cross-reference information from multiple sources in ways that were previously impractical. For example, authorities can analyse publicly available social media data to identify inconsistencies between an individual’s reported income and the lifestyle they portray online. The Swedish Tax Agency has already demonstrated the effectiveness of this approach by using already available data to notice patterns and make predictions to target potentially incorrect or fraudulent commuting deduction claims. In its first year alone, the initiative reportedly generated over $4 million in additional tax revenue. To respond to this changing environment, taxpayers themselves can use AI tools to perform “mock audits” on their own data, helping them identify potential risk areas before tax authorities do.
Like many other business functions, tax departments are increasingly embedding AI into their compliance and reporting processes to eliminate repetitive manual work and allow teams to focus on higher-value judgement-based activities. AI solutions can already:
- review and code purchase invoices for indirect tax compliance,
- categorise expense data for tax computations,
- analyse and translate foreign-language tax returns,
- draft technical memoranda,
- support transfer pricing analyses by reviewing intercompany transaction data and benchmarking pricing across jurisdictions, and
- model and manage the significant data requirements arising from OECD Pillar Two global minimum tax rules.
Whilst the uses are numerous, as with any automation initiative, the quality and structure of the underlying data is critical. Businesses investing early in modernising, organising, and standardising their data environments will be best positioned to benefit from AI within the tax function.
AI tools are also increasingly capable of providing instant tax advice. While this may be suitable for lower-risk scenarios, it raises an important question: since an AI tool cannot hold legal liability, who bears the burden when advice is incorrect? The protection and accountability that professional advisory firms provide will continue to carry significant value, perhaps making it even more important for businesses to scrutinise liability caps and engagement terms carefully. There is also the danger of overreliance on AI-generated output without properly verifying the underlying sources or technical accuracy. The consequences of mistakes may not become apparent until years later during a tax audit or dispute. Tax legislation also evolves constantly, and AI tools may not always reflect the latest law, case precedent, or regulatory guidance.
Where I personally struggle to see AI having a transformative impact is in the delivery of high-value strategic tax advice itself. For smaller or lower-risk matters, I would feel comfortable relying on an AI-generated response. But where millions of dollars are at stake, experience, judgement, and commercial awareness remain irreplaceable.
This reflects a broader trend emerging across many industries: AI is likely to replace much of the repetitive junior-level work, while experienced professionals remain essential. However, this raises a longer-term challenge for the profession. If junior professionals no longer gain experience through the traditional repetitive processes that historically formed the foundation of their training, how do we develop the next generation of senior tax advisors in 10, 20, or 30 years’ time? The answer may be that junior professionals need exposure to higher-level strategic discussions much earlier in their careers. Developing judgement, critical thinking, and commercial awareness may become far more important than mastering purely mechanical processes.
There are also important cybersecurity and confidentiality concerns. Tax data is among the most commercially sensitive information a business holds, and organisations will need to carefully consider what information is being input into AI systems, where that data is stored, and how it is protected. Businesses will also need robust governance frameworks around the use of AI in the tax function, particularly where AI-generated outputs are used to support tax filings or technical positions. Being able to explain how an AI tool arrived at a conclusion, and demonstrating appropriate human oversight and review, may become increasingly important during audits or disputes with tax authorities. For that reason, human review by appropriately experienced professionals will remain essential.
AI is set to fundamentally reshape the tax profession, particularly in compliance, data analysis, and process efficiency. While it has the potential to reduce administrative burden and improve the quality of insights available to both taxpayers and tax authorities, it is unlikely to eliminate the need for experienced human judgement. The firms and tax functions that succeed will be those that combine strong technical expertise with high-quality data, effective governance, and a thoughtful approach to integrating AI into existing processes.
