Two-thirds of finance professionals surveyed for Parliament Street expressed reservations about large-scale artificial intelligence projects despite growing pressure to adopt the technology. The think tank commissioned research among 500 chief financial officers and finance professionals at medium-sized and large accounting businesses, with 67% reporting concerns about major deployments. The findings highlight tensions between efficiency ambitions and the confidentiality, accuracy and regulatory obligations of accountancy work.
The most frequently cited areas of concern were data leakage, mistakes in AI-generated work and the possibility of compliance breaches. These risks are particularly relevant to accountancy practices handling confidential customer records, statutory reporting and financial decisions. An automated tool may produce a plausible answer quickly, but staff still need to know which source documents were consulted, whether the calculation is correct and who is responsible for approving the result.
Parliament Street also found that 80% of its respondents said they would use AI to reduce costs or improve processes. The figures do not demonstrate a straightforward division between advocates and opponents. A person may support carefully controlled automation and remain concerned about deploying it across sensitive operations without sufficient testing. The distinction between willingness to experiment and confidence in governance is important when interpreting the results.
Parliament Street reports that 57% of respondents said their organisations lacked the in-house skills needed to manage deployments effectively. Some respondents expected to address this by recruiting graduates with AI-related skills, while others planned internal training. Technical knowledge alone is not sufficient for every accountancy use case: employees also need judgement about materiality, professional confidentiality, model limitations and the consequences of approving incorrect output.
Parliament Street also reports that 72% of its survey respondents expected their AI budgets to increase in 2027. Budget intentions are not audited spending or confirmation that planned projects will proceed, but they suggest organisations are considering further investment even while expressing unease. The study also reported that 55% were considering redundancies associated with AI in the following year. Consideration is not a completed workforce decision, and the survey does not establish the number of jobs that might ultimately be affected.
For accounting firms, governance starts with defining acceptable uses. Summarising a public regulation is different from analysing a client’s payroll records or preparing a document used in an audit. A useful operating model can categorise tasks by sensitivity, require review for high-risk outputs and prevent data from being sent to services that do not meet confidentiality or security requirements. Clear responsibility is needed even if the first draft is generated automatically.
Errors pose a more subtle risk than a visibly failed system. A fluent but incorrect explanation may enter a working paper and be repeated without challenge. Firms therefore need checks designed around the particular task, such as reconciling figures to original ledgers or preserving an audit trail of data and assumptions. Conventional quality control remains necessary; AI can change the speed and volume of work without removing responsibility for professional standards.
Workforce planning introduces a separate challenge. If junior employees no longer perform some routine reconciliation or document-preparation tasks, firms must ensure they still develop the underlying expertise required to supervise automated processes later in their careers. Recruitment of AI specialists may be useful, but it does not replace technical accountancy training. Organisations will need to design roles that combine new tools with professional learning rather than assuming one skill set substitutes for the other.
The research concerns selected medium and large accounting practices, so its percentages cannot automatically be extended to all UK accountants or every small bookkeeping business. It also relies on self-reported attitudes rather than direct measurement of breach rates, project costs or savings achieved. The study helps identify the issues practitioners say are influencing their decisions, but does not demonstrate that a particular approach to AI governance is effective.
Parliament Street’s survey found both reservations and interest in AI among accountancy professionals: 67% expressed concerns about major deployments, while 80% said they would use the technology to improve processes or reduce costs. The same group identified skills shortages and issues of confidentiality and compliance. The results cover respondents at medium-sized and large accounting businesses, where the practical challenge is to combine automated work with professional checking and accountability.
You must be logged in to post a comment.