Non-normative guidance focuses on supporting innovation while maintaining appropriate standards, according to regulators.
The initial guidance on the use of artificial intelligence in the Financial Reporting Council (FRC) audit is welcomed as the first step towards clarity in regulations regarding the adoption of artificial intelligence (AI), but further clarity is needed.
As the use of AI tools in audits grows, FRC's new guidance outlines a consistent approach to implementing hypothetical AI-enabled tools and provides insight into FRC document requirements that Regulators are designed to support innovation across audit experts.
ISQM (UK)1 requires companies to acquire, develop, implement, maintain and use appropriate technical and intellectual resources to enable operation of quality control systems and engagement performance, documenting how they addressed risks that did not meet these objectives.
Support Auditor
This guidance aims to support auditors and central teams of auditing companies in developing and using AI tools in their work, while also providing third-party technology providers with regulatory expectations for their customer base.
The intended scope of guidance spans deep learning models, including traditional machine learning techniques and generation AI. It was developed in collaboration with FRC's Technology Working Group and is based on technical experts across audit experts.
The main features of the guidance are:
- Two-part structure – An example of one potential way that AI can be used in audits aimed at supporting proportional and robust documentation of tools that use AI.
- A sophisticated view of appropriate explanationability – Accept that the appropriate level of explanability differs depending on the context and usage.
- Cooperation with government AI principles – Document guidance reflects the five AI principles of the UK government.
- Market-wide related – Guidance includes material that clarify how expectations are translated into the context in which the tool is retrieved from a third party.
Mark Babington, FRC Executive Director of Regulatory Standards, said: “AI tools are moving beyond experimentation and into reality in specific audit scenarios. Responsibly deploying them can improve audit quality, support market trust, promote innovation and ultimately contribute to the UK's economic growth.
“This guidance aims not only to clarify the FRC's expectations regarding proportionally appropriate documentation of tools that use AI, but also to explain how AI can enhance auditing work. This area is moving quickly and we recognize that it supports innovation and proper use across the profession, both in the UK and internationally.”
Uncertainty as a barrier to AI adoption
Esther Malova, ICAEW's head of technology policy, said regulatory uncertainty is one of the barriers to the meaningful and large-scale adoption of AI, urging ICAEW to seek clarity in its white paper science, innovation and technology consultations to AI regulation. This approach advocates sector-driven regulations for AI, and sets out the requirements for existing regulators to use AI within their domain.
Ian Pay, head of data analysis and technology at ICAEW, says this latest FRC guidance represents a positive first step towards providing some clear companies that are needed by the clear companies that are needed in response to regulatory expectations when it comes to corporate AI adoption. However, there is a lot more to do.
“We value FRC's contributions in this area, and we know that conversations with our members and companies will appreciate guidance on the use of AI in audits. We also know that many companies are working on the details of AI implementation and the desire for practical examples and use cases of AI.
“We hope to work with the FRC in the coming months to develop this and make it even clearer. Maintaining AI in audits is important in supporting businesses of all sizes, as we strive to carefully balance innovation, risk and ethical adoption,” Pay said.
Supports innovation
The FRC Guidance also demonstrates support for innovation across the auditing department. This is what the regulators publicly state as an ambition in line with the government's mandate towards UK regulators to support growth and innovation throughout the UK economy.
In addition to the guidance, FRC has also published a thematic review of the processes of six large companies to prove the new technologies used in audits. This includes insights and examples of excellent practices in the processes and controls to authenticate the automated tools and techniques used in audits.
Companies other than the six-person focused on reviews will find much of the content interesting, but you may feel that much of what they do is out of their reach. However, thematic reviews show that many of the principles regarding the documentation of AI tools are applied more widely, so that even companies that are not currently researching AI tools themselves can receive learning from publications.
Through guidance, the possibility of independent guarantees for third-party AI models is mentioned. Malova says obtaining this guarantee could be challenging in the relatively early sector.
“There are many types of activities labeled as AI assurance and auditing companies. You need to ensure that assurances gained through the AI model cover the associated risks and achieve the intended purpose. Furthermore, many companies that use generative AI to support audit testing have confirmed that access to development and testing information is not yet available.
She continues: “Companies also need to understand where the responsibilities of third-party models end and where responsibilities as implementers and users begin. ICAEW will actively tackle these challenges and help promote an effective AI assurance ecosystem that will help members and the wider society to be more sensitive to AI.”
Franki Hackett, Grant Thornton's Head of Audit AI, said he was pleased that the company contributed to FRC's regulatory guidance on the use of AI in audits. “This guidance provides much needed clarity on how to respond to the current state of AI without constraining innovation as technology is developed. As audit quality relies on good technology, we incorporate this guidance into the AI quality framework to ensure that AI research and development delivers optimal results for our clients.”
ICAEW recently held its first AI Assurance Meeting, bringing together players and experts from key industry companies to discuss a variety of topics, including what actually is AI Assurance? Get more guarantee than the basic model.
