How eagle-eye AI is transforming expense management
AI has been helping the finance sector to streamline and optimise processes for some time now, but the technology has never been more relevant or useful to finance chiefs than now. Remote work is here to stay, and finance teams are increasingly looking at alternative ways to make their expense management processes more structured, accurate and efficient. The current pandemic has only highlighted the challenges that organisations face during the spend auditing process. These issues range from the slow pace of manual expense processing, to the need for updated company policies.
AI can catch what the human eye misses
Expense audit has historically been plagued by out-of-date and ineffective technologies, on top of a laborious manual process. In contrast, AI-powered spend auditing enables finance leaders to take a more proactive, precise approach to expense reimbursement. AI is capable of auditing 100 per cent of expense claims and is far more nimble at spotting errors and duplicate payments, as well as flagging questionable spend.
Below are some notable instances of when AI has flagged discrepancies in expense reports submitted by employees that would otherwise have gone undetected.
“Airbnb” – on a mate’s sofa
An employee stayed at his friend’s home whenever he was in town on business. Not a problem, except that he and his friend hatched a plan to run the expense through as an Airbnb stay. The friend even went so far as to post the home on Airbnb at a ridiculously high price, out of line with similar accommodation in the area.
The employee then submitted the inflated-price Airbnb stay in their expenses. Once reimbursed by his employer for the expense, the friends then split the proceeds between them. However, AI immediately detected the higher-than-average room rate for the style of accommodation and location. This was then flagged to the auditor for further review, stopping the schemers in their tracks.
Just a free lunch - or a backhander?
When it comes to employee expenses, if an employee commits bribery, it’s not a defence for a business to claim that it didn’t know about it. Under the UK’s Bribery Act, lunches, gifts, or “anything of value” can be considered bribes, and this can result in astronomical fines and untold damage to an organisation’s reputation.
Manual expense reviews are unlikely to spot the names of foreign officials or heads of state associated with meal receipts, but AI can easily detect them. This ensures that if employees are dining with high-risk individuals, the expense can be further reviewed to see if it’s unreasonably extravagant – and keeping a company out of regulatory hot water.
Much like the euphemistic moniker, “gentlemen’s club,” strip clubs like the well-known Spearmint Rhino often have a ‘doing business as’ (DBA) name that differs from the establishment’s name. This DBA, or ‘trading as,’ name appears on receipts so the line item can avoid attracting the attention of expense report approvers.
AI is trained to recognise that “K-Kel, Inc” is not in fact the name of a restaurant, or other approved venue, but is instead the ‘trading as’ name for Spearmint Rhino. It really is surprising how often this scenario comes up, but AI is trained to identify these and other instances where DBA names are masking potentially inappropriate expenses.
Although these expense red flags may seem outrageous, or even obvious in retrospect, given the volume and depth of many organisations’ expense claims, manual review simply can’t compete with the power of AI. AI ingests data from a seemingly limitless amount of online resources, in real time, and continuously builds intelligence that enables it to identify problematic spend – something that human auditors simply don’t have the time or resources to do.
To mitigate these spend risks, businesses can streamline the spend audit process with the help of AI. Having technology in place that automatically approves low and medium-risk expense reports allows auditors to focus their precious time on the high-risk items that matter the most – preventing fraud, reducing spend, and ensuring regulatory compliance. AI is crucial to the modern expense auditing process, and is a technology investment well worth investing in to save time and money now, and in the long run.
By Andrew Foster, VP Consulting, AppZen
Facebook Develops AI to Crackdown on Deepfakes
In light of the large tidal wave of increasingly believable deepfake images and videos that have been hitting the feeds of every major social media and news outlet in recent years, global organisations have started to consider the risk factor behind them. While the majority of deepfakes are created purely for amusement, their increasing sophistication is leading to a very simple question: What happens when a deepfake is produced not for amusement, but for malicious intent on a grander scale?
Yesterday, Facebook revealed that it was also concerned by that very question and that it had decided to take a stand against deepfakes. In partnership with Michigan State University, the social media giant presented “a research method of detecting and attributing deepfakes that relies on reverse engineering from a single AI-generated image to the generative model used to produce it.”
The promise is that Facebook’s method will facilitate deepfake detection and tracing in real-world settings, where the deepfake image itself is often the only information detectors have to work with.
Why Reverse Engineering?
Right now, researchers identify deepfakes through two primary methods: detection, which distinguishes between real and deepfake images, and image attribution, which identifies whether the image was generated using one of the AI’s training models. But generative photo techniques have advanced in scale and sophistication over the past few years, and the old strategies are no longer sufficient.
First, there are only so many images presented in AI training. If the deepfake was generated by an unknown, alternative model, even artificial intelligence won’t be able to spot it—at least, until now. Reverse engineering, common practice in machine learning (ML), can uncover unique patterns left by the generating model, regardless of whether it was included in the AI’s training set. This helps discover coordinated deepfake attacks or other instances in which multiple deepfakes come from the same source.
How It Works
Before we could use deep learning to generate images, criminals and other ill-intentioned actors had a limited amount of options. Cameras only had so many tools at their disposal, and most researchers could easily identify certain makes and models. But deep learning has ushered in an age of endless options, and as a result, it’s grown increasingly difficult to identify deepfakes.
To counteract this, Facebook ran deepfakes through a fingerprint estimation network (FEN) to estimate some of their details. Fingerprints are essentially patterns left on an image due to manufacturing imperfections, and they help identify where the image came from. By evaluating the fingerprint magnitude, repetition frequency, and symmetrical frequency, Facebook then applied those constraints to predict the model’s hyperparameters.
What are hyperparameters? If you imagine a generative model as a car, hyperparameters are similar to the engine components: certain properties that distinguish your fancy automobile from others on the market. ‘Our reverse engineering technique is somewhat like recognising [the engine] components of a car based on how it sounds’, Facebook explained, ‘even if this is a new car we’ve never heard of before’.
What Did They Find?
‘On standard benchmarks, we get state-of-the-art results’, said Facebook research lead Tal Hassner. Facebook added that the fingerprint estimation network (FEN) method can be used for not only model parsing, but detection and image attribution. While this research is the first of its kind, making it difficult to assess the results, the future looks promising.
Facebook’s AI will introduce model parsing for real-world applications, increasing our understanding of deepfake detection. As cybersecurity attacks proliferate, and generative AI falls into the hands of those who would do us harm, this method could help the ‘good guys’ stay one step ahead. As Hassner explained: ‘This is a cat-and-mouse game, and it continues to be a cat-and-mouse game’.