Using AI to Combat Fraud Risk

By Andrew Foster, VP consulting, AppZen

Fraud experts use three factors to explain the motivation for an individual to commit fraud – pressure, opportunity and rationalisation. While the percentage of employees committing fraud is very low, this model, known as the Fraud Triangle, is used to explain what lies behind it for those who do commit it.

Coronavirus has had a huge impact in driving up fraud levels because it has increased all three elements of the Fraud Triangle and has pushed the risk of financial fraud to unprecedented levels.

Many people are clearly now under increased financial pressure. Coupled with this, companies have changed working practices, and cut or furloughed staff, putting traditional controls under intense pressure and creating more opportunities for those few individuals who are now more able to rationalise fraud.

Combatting rising fraud risk

Finance professionals and business managers are clearly facing many challenges in the current climate and fraud mitigation may not be the top priority. Unfortunately, attempted business spend fraud is increasing and will go up in the coming months and years. By taking proactive steps and implementing the right approach it is possible to protect company expenditure and also to save time and money.

At a time when costs are under great strain and adding investigative staff is a hard sell, how can organisations stop would-be fraudsters? It turns out there is some good news. By rethinking controls regimes and mining the data already coursing through existing enterprise systems, businesses can identify, investigate, and remediate potential fraud – and even prevent it in the first place.

AI-powered software developed specifically for finance teams can revolutionise spend and expense management, enabling organisations to quickly and accurately identify spend risks and combat fraud. This approach helps companies rely less on an auditor’s luck in catching a few bad actors and more on a systematic, data-driven and fair approach.

There are four key reasons why AI understands financial documents better than anyone else:

1. AI looks at everything: every transaction, every line item.
2. AI remembers everything, so it can spot anomalies and duplicates over time.
3. AI doesn’t just look at the transactions themselves, but at their context.
4. AI is continuously learning from the collective experience, insight, and problem spots from their large
   

   Andrew Foster

   enterprise customers.

Analysing huge volumes of information without an army of analysts

AI software reads financial documents – like POs, receipts, and contracts – and extracts meaning from all the structured and unstructured data in those documents to identify the problem areas. Because AI can process such a huge volume of transactions, it’s extremely accurate at recognising when something is awry. Most of the time, it isn’t fraud – it’s a duplicate receipt, or someone expensing an out-of-policy item. Regardless, AI systems save finance teams’ time by highlighting only high risk transactions that need review and eliminating the need to spend time spot checking transactions (the majority of which are just fine).

Perfect memory

AI systems remember every transaction and document ever processed, which makes catching duplicates over time, across departments, or price or quantity anomalies, straightforward and easy. They will systematically identify which employees are over the policy line; for example, which employees are consistently spending slightly higher than the stipulated meal limit, delivering an objective way to see who is pushing the limit of T&E policy, not being diligent, or even committing fraud.

Using collective market insight

AI software vendors have large numbers of companies using their systems and by collecting information across all their customers they have unrivalled insight into a lot of different kinds of misbehaviour. This enables them to spot trends or common fraudulent behaviours. As their systems learn, they share patterns that represent high risk (without sharing the transaction data) so that all users of their systems can benefit from what they’ve seen and learned from the entire customer base. This increases their effectiveness and accuracy in identifying and stopping expenses fraud.

Customisable controls

Of course, in many cases, what’s out-of-policy for one company is in line with policy at another. AI systems combine the benefits of scale with individual controls that let each customer automate their own policies and tailor the risks they care about most by creating custom rules to build their own spend policies, controls and risk levels.

Conclusion

Through an automated, AI-driven approach, finance teams can catch duplicates, stop wasteful or over-the-limit spend, identify data entry typos, find incorrect receipts and block out-of-policy expenses, far more quickly and effectively than any manual process ever could.

In these times where resources are being stretched to the limit and fraud risk is at a record high, streamlining the process and removing error-prone manual tasks gives finance teams far better visibility into patterns of behaviour and spend within an organisation, and eliminates wasteful, fraudulent or out-of-policy spend.