Chamber and committees

I am sure that, when the Auditor General comes knocking on the Scottish Commission for Public Audit’s door for more money, that will be part of his pitch.

I have one final question. Are you doing anything in relation to Covid-related fraud? Obviously, there has been a lot of noise around the potential scale, volume and value of many different aspects of Covid spending, particularly around the work of HMRC in relation to loans, grants and so on, but there may be other bodies that you do work for that have been affected by Covid fraud, to use that phrase.

I presume, then, that the media article that I came across in my research ahead of today’s meeting was erroneous in claiming that Perth and Kinross Council is

“one of ... two UK local authorities which does not share ... electoral roll”

data

“with the National Fraud Initiative”.

Is that true, or is the article incorrect?

Okay. Perhaps that is something that we can follow up with the Convention of Scottish Local Authorities, for example, as the body that assists and represents a number of local authorities in Scotland.

Do you think that the industry is a bit behind the curve in that respect? Other people are already making extensive use of AI in their business processes.

Okay. Would there be any reason why it would not? I am not picking on it—it is just that it has been flagged as the only council in Scotland that does not participate.

I have two slightly random questions. With the advancement of the ability to read and calculate data using artificial intelligence, have you seen much of a move towards AI over the past five years?

We have often talked about that side of audit and the potential benefits of using AI to speed up processes and improve the quality of analysis—of course, we have also talked about the potential for effectively putting nice, well-paid auditors out of work. Are there any potential advantages to the further use of AI in the work that you do with regard to identifying large amounts of data from nearly 120 public bodies, all of which submit data in different forms and formats, using different technologies?

It is not always fraudulent activity that you flag up. For example, I presume that bodies that participate in the NFI submit data sets and your analysis flag up issues such as duplicate payments. The report says that around £750,000-worth of duplicate creditor payments were identified. Do you know how much of that was erroneous, or what number of payments were identified as duplicate matches of data but were actually valid? For example, as you say, you could buy a £2,000 laptop from a creditor and the next day do exactly the same thing again. That would perhaps be flagged as an erroneous payment and potential fraud, but it could be a valid payment made by a body. What work do you do thereafter to match or marry up the value in that data identification with what happened next or the follow-up activity?

That would be good—thank you.

I have two quick questions. Housing benefits and pensions are quite big pieces of work as part of the initiative. I perhaps did not understand some of the terminology, as there is a lot of jargon in the report. Talk me through what it means when we are told that the average individual value of overpayments for housing benefit rose from just over £2,000 to £6,500 in a short period. Is that to do with the value of payments or with the level of potential fraud?

I appreciate that, since 2018, there has been a marked shift from a bespoke housing benefit payment, for example, to universal credit, which has perhaps mopped up some of the more individual benefits. However, I could not quite get my head around the situation with housing benefit fraud, such as whether you have identified a rise or decline.

If fraudulent housing benefit claims are picked up through the work that you do, do you tell the DWP or the local authority about that?

They all do.