Should accountants be worried about AI this tax season?

Over half of UK taxpayers plan to use AI to complete their self-assessment tax return. Do accountants need to be concerned about how their clients are using it?

A TaxFix study found that 59% of UK taxpayers are using AI to help them complete their self-assessment tax return for HMRC’s 31 January 2026 deadline. The study also asked accountants about their concerns, which include AI’s incomplete advice (73%), inability to handle nuance (62%) and its lack of UK-specific information (52%).

Are accountants seeing clients relying more on AI for managing their tax affairs, or turning up to appointments armed with AI-generated information? What is the view among UK accountants of the impact of AI on their work and their client interactions?

We welcome clients doing research themselves

Simon Antonis, Founder and Director, Simon Maughan & Co

AI is giving much more detailed and customised responses than search engines, which most clients have been using for many years. But AI often misses important areas that are relevant to their situation, and unless you know what questions to ask to challenge AI, those gaps stay hidden. AI can’t always reference all the relevant regulations, latest practices, or legal precedents, and it definitely can’t apply these to specific circumstances.

But we actually welcome it when clients do their research, whether through AI, search engines, or chatting with friends and family. We can dive straight into providing tailored advice while making sure everything stays within the law. This research process often helps clients see these limitations and complexities for themselves.

AI is a useful tool for summarising research and great for marketing – and our website development team uses it too. The key thing clients often discover is that tax advice isn’t just about knowing the rules. It’s about understanding how those rules interact with your specific situation, your goals, and sometimes even your industry. That’s where expertise and experience really count.

Verdict: We welcome the client interactions AI creates; it shows where expertise and experience really count.

It’s our job to verify information for clients

Stephanie Marshall, author and CEO, A B C S UK

As accountants, we’ve heard “my mate Dave down the pub says…” for years. Generative AI tools are simply the modern version of Dave down the pub or a quick Google. I haven’t seen an increase in these questions, and I suspect those asking generative AI for tax advice often do not have an accountant to support them.

Accountants don’t need to fear generative AI. Due to the unique nature of each taxpayer and company’s circumstances, generative AI would need to be far more sophisticated before it could accurately provide tax advice correctly and confidently.

If clients are bringing AI-generated information to a meeting, it’s our job as accountants to verify it. The relationship built with a client means they should trust your expertise over AI every time.

AI is great for pulling together information such as current HMRC guidance, tax and VAT thresholds and allowable expenses, as a useful starting point. But as personal circumstances and business structures vary, applying tax advice to the client’s position is where a qualified human accountant is needed.

Verdict: There’s no need to fear AI; the human touch is still essential, and relationship-building should mean clients trust your expertise.

No substitute for understanding a client’s situation or the finer points of tax law

Sean Farnell, Partner, Burgis & Bullock

We’re starting to see clients arrive with information they’ve picked up from AI tools, though in truth, most still turn to Google for quick answers. For straightforward questions, AI can sometimes provide a decent steer. But as soon as things get more complicated, or there are grey areas in the legislation, the answers can be wide of the mark. There’s no substitute for understanding a client’s situation or the finer points of tax law.

We’re only just starting to use AI in the firm, and we’re very careful about how and where it’s applied. Confidentiality is non-negotiable, and everything is subject to proper review. AI is helpful for sifting through large volumes of data and spotting patterns, but it can’t interpret context or apply judgement in the way a professional can.

AI isn’t going away. The fact that more people have access to information is, on balance, a good thing. But there’s a real risk in relying on generic or unregulated advice, so professional standards and regulated advice matter more than ever.

The best results will come from accountants and AI working together, using technology to improve efficiency and insight, anchored by professional expertise and a proper understanding of each client’s needs. Humanising the numbers will increasingly be what clients are willing to pay for, with soft skills increasingly at a premium.

Verdict: AI can’t replace accountants, and best results will come from us working with technology.

Models built on verified, expert-curated data are the real game-changers

Ben Chaplin, Chief Financial Officer, Peninsula Group

Accountants and professional bodies are watching developments with interest and considering how best to harness the technology to improve efficiency, balanced with risk. Using off-the-shelf AI can introduce serious risks such as outdated or incorrect data, fabricated – hallucinated – answers and unverified output.

Principles set out in our ethical and auditing standards stand more than ever for finance professionals. Apply analytical rigour and professional scepticism, consult appropriately and ultimately always validate the source of any information. As with any technological development, different forms of AI are ultimately just another tool we can choose to use, subject to compliance with professional standards. All of us need to consider the sustainability impact and mitigate the risks to protect ourselves and our clients.

The real game-changer is models built on verified, expert-curated data, rather than the fastest internet scrape. We use a solution that leverages decades of expert-written tax, audit and accounting content, providing responses with links to statutes, case law and detailed analysis, ensuring answers are accurate, reliable and verifiable.

Verdict: Seek out the best models, be rigorous and professionally sceptical and always validate AI-generated information.

Accountants need to be able to explain why AI’s information is wrong

Arj Kumar, Founder, Taxd

Typically, accountants build relationships with clients and learn a lot about them. This gives them unparalleled insights into their long-term goals, priorities and financial situation. Unlike AI, accountants’ view of their clients’ money is often shaped by context and by years of relationship building.

This isn’t something which AI can do to a satisfactory level. However, this may be possible in future. As it stands, AI can effectively be used for straightforward questions, such as ‘how to invest £10,000 every year’.

But for long-term financial planning and complex situations, traditional financial advisors will be preferred by the majority. For trained accountants, it’s frustrating to see clients using AI, receiving incorrect information and treating this advice as gospel. It’s important for advisers to ensure their clients are aware of the risks.

Now, many clients come equipped with advice from GenAI. Although this can be a good start, this sometimes can lead to wrong information being shared. To counter potential mistakes, accountants need to be able to explain why AI can be wrong and back up their responses with evidence. This has fundamentally flipped the relationship between accountants and clients.

AI can be used to provide confidence or as extra assurance. In most cases where the advice needed isn’t too complicated, AI can draft things up and get 99% of what the recommendation would be. This can be reviewed by a human to remove any errors or unhelpful advice.

If firms move fully to AI for advice, they will need to consider insurance to cover liabilities that could arise if inappropriate AI-driven responses are acted on. There is hesitation, but we are seeing more uptake in AI adoption as providers look to remain competitive.

Verdict: Unlike AI, accountants’ view of client money is shaped by context and relationship building.

Georgia Lewis is a journalist who has worked in Australia, the Middle East and the UK. Over 30 years, Georgia has covered a diverse range of subjects and industries, including business, insurance, technology and logistics..