By Cat Hall Artificial intelligence Artificial Intelligence 101 3 Aug 2026 Your overview of the kinds of AI available, its risks, and guidance on how to use it effectively in financial services. Artificial Intelligence (AI) has been broadly adopted by accountants and bookkeepers. A 2025 survey of UK practitioners by professional services firm Wolters Kluwer found that 91% were either using some form of the technology or were planning to do so in the near future. There’s a lot of noise about the technology, so it can be difficult to unpick which kind of AI might be best suited to the task you’re working on. This overview considers: The kinds of AI available, and what they are best for The risks of using AI Official guidance on its use in financial services How to use AI effectively The effect it’s having on the industry. Types of AI Agentic Systems that don’t need as much supervision as other forms of AI (hence the name). These models have agency to act, and aren’t limited to the information in their training datasets: they can access the internet and query databases, and then make informed decisions. They can also work across multiple steps of a problem, and work to a longer-term horizon. This kind of AI is set up to achieve tasks. Accountants can use it to automate workflows, including processing invoices. Examples include QuickBooks and FloQast. Generative Models react to prompts from users to generate content. They’re limited to the information contained in the large datasets they’re trained on, training which lets them identify patterns and relationships within the data so they can predict which word or phrase generally comes next in certain contexts. However, they do not fundamentally understand facts and meaning in the same way as people do. Accountants can use it for routine writing, such as emails. Examples include ChatGPT and Copilot. Predictive Identifies patterns in historical data, and uses them to forecast future outcomes. Accountants can use it to forecast cash flow, predict revenue and expenditure trends and support budgeting and financial planning. Examples include Xero Analytics Plus and Sage Intacct. Risks of AI AI does come with downsides, including data leaks and GDPR violations. In 2023, employees at Samsung’s semi-conductor business leaked corporate secrets by keying proprietary code into ChatGPT. It’s not necessarily immediately obvious why inputting data would be a problem, so it’s worth bearing in mind that AI tools often process information through external servers. This means confidential information can become part of system records if not carefully handled – as in this case. It can also ‘hallucinate’, or make up information in response to prompts. Deloitte provided a high-profile example of this when it published an ‘error-ridden’ report in 2025, including made up footnotes and references. The Big Four firm took a reputational and financial hit, having to partially refund the Australian government. This means a human subject-area expert should always check AI’s output. There’s also the issue of human bias in the training data or algorithm. Where preconceptions and assumptions are built in, outputs can be unreliable and incorrect, and can lead to reputational damage. Users should also beware relying too heavily on AI. It’s easy to miss hallucinations and misinformation, so a human expert should always review output before making use of it. We discussed how accountants in practice are managing AI risks here. AI e-Learning Members can dive into our bitesize CPD to learn more about privacy and compliance risks, and generative AI for accounting tasks. Learn more Guidance and regulations around AI As a result of these problems, regulatory bodies are paying attention. The Financial Reporting Council (FRC) published its first guidance on the use of AI in audit in June this year. It requires firms performing audits to use appropriate technical resources for quality management, and to document how they address risks. The guidance isn’t prescriptive, but looks to clarify expectations. AAT members can read analysis on the guidance here. Professional Conduct in Relation to Taxation (PCRT) bodies have published guidance to help members apply PCRT fundamental principles when using AI tools in tax work. The guidance discusses examples of how AI tools could be used in tax, and identifies possible ethical risks, and the safeguards to implement. A key takeaway from this document is that ‘outputs from AI tools should not be used as authoritative tax or legal advice, with reviews to be undertaken by a qualified professional in the specific context of the client to whom the advice is being provided’. How to use AI effectively In a conversation with AAT, John Toon MAAT, Technology Strategy Lead at Beever and Struthers, explained what people get wrong with AI, and how to approach it. “The AI conversation is now all about large language models (LLMs) and fun stuff like ChatGPT and Gemini. But the reality is that accounting practices and businesses in general struggle to find use cases. And that’s down to a number of reasons, but mainly it’s because a lot of firms have thought, ‘I’m not really sure what it does, so I’m going to experiment with it and just spread it around everywhere and let 50 people go away and play with this with no real structure around what they’re trying to achieve.’” Instead, Toon says businesses that focus on problem-solving have much more success in utilising LLMs. He cites the team’s creation of a proof of concept of effectively an LLM-powered chatbot to do personal tax returns. Read more of this conversation here. Changing the industry Some people, such as Jeremy Hunt, have suggested AI could undermine the industry. Many others, including AAT, see its potential to automate fiddly and repetitive tasks. AI has the potential to revolutionise the industry, but most likely this will be through freeing up accountants’ time to work on more strategic areas. One potential outcome is that more people will move into the industry, addressing the skills gap. According to AAT’s 2025 research, two in five people would consider a career change to accountancy if administrative tasks were carried out by AI. Additionally, one in five accountants who have left the profession said they would return if automation could remove the more administrative aspects of the job – and 23% said they would have stayed in the profession longer if AI had been available to help them. Read more about why AI is making accountancy exciting for a new generation and how AI is transforming accountancy’s appeal on AAT Comment. Of course there are some concerns about what the rise of AI means for entry level roles, with the Big Four cutting graduate jobs. Firms must be careful not to curb entry-level intake too heavily or they risk restricting the talent pipeline even further. Accountants discuss how firms should redesign junior career development here. Tech and AI essentials for finance Understand how AI is shaping finance, and discover how to use it in your day to day work with confidence. Find out more Cat Hall is Content Specialist at AAT.