Should businesses brace for AI reviewing their accounts?

AI is seeping into every corner of business operations, and many firms are already deploying tools to automate routine tasks.

Now it looks as though regulators may be set to follow suit as reports suggest the Bank of England is exploring AI-driven inspections to trawl the vast volumes of financial data it and other regulators hold.

If automated scrutiny becomes mainstream, it will change how compliance is enforced and businesses should be ready to adapt to any changes this brings.

How might AI-powered checks affect firms?

HM Revenue & Customs (HMRC) already employs automation to support investigations, but the scale and scope of its AI use remain opaque.

A recent tribunal forced HMRC to disclose how it deploys AI when assessing R&D tax-relief claims, amid concern that off-the-shelf models may not be suited to complex decision-making.

This debate underlines a wider issue surrounding the use of AI as only specialist, tailored AI is capable of providing the long-term value sought by HMRC and the Bank of England.

Specialist AI, built for narrow tasks and used by trained teams, can be extremely effective as it removes tedious work, prioritises cases for human attention and surfaces anomalies that warrant further enquiry.

By contrast, generic models can hallucinate, misinterpret context or generate false information.

These problems are bad whenever the AI is used but become truly unacceptable when financial positions and reputations are at stake.

The Bank of England’s interest therefore raises a key question of whether the AI by tailor made and whether the staff working with it will receive specialist training.

The hesitation on the part of HMRC to divulge its use of AI has caused many to become suspicious that the AI may not be entirely suited to the complexity of the task at hand.

If that is the case, the question of the security of the system becomes more pressing.

Some AI systems remain vulnerable to prompt injection where maliciously crafted inputs override prior instructions or influence behaviour.

Although some effort has gone into reducing the scale of this problem, it is not something that AI is entirely free from as of yet.

The use of vulnerable AI systems has prompted warnings that sophisticated actors might attempt to encode filings in ways that confuse or bypass automated checks.

If reviews rely more heavily on models, regulators will need strong safeguards and businesses should assume automated tools will be probed for weaknesses.

How can businesses prepare for AI-driven financial reviews?

Whether AI is already being used at scale or is still under consideration, it is best to act as though your filings are already being reviewed by robots.

This could mean that there are more rigorous checks being conducted on any of the filings you are currently submitting so it may be time to think about how accurate these filings are.

Seeking professional help and support is the most effective way of catching errors before they become larger problems.

We are on hand to help you keep up with the latest regulations and requirements while ensuring that you finances are suited to the needs of your business.

As we learn more about how AI is being used to monitor filings, we will keep you updated.

To keep pace with more rigorous financial reviews, speak to our team today!