How Can AI and Machine Learning Take Fraud Detection To The Next Level?
Jul 27, 20205 min read
Marketing Automation Specialist at Ideamotive
The history of frauds is probably as ancient as the history of money themselves, or maybe even older. With the era of digital payments and internet trade, both frauds and anti-frauds entered a new dimension, where human minds can be insufficient, and artificial intelligence shines.
When thinking about frauds, one usually recalls a Ponzi scheme, an evergreen trick to attract the money of over-greedy and careless investors. But fraud is a part of every business in probably all time periods. Just think about Archimedes - yep, the famed ancient Greek mathematician, physicist, engineer, and astronomer.
A famous anecdote about him states that he got the idea of floating bodies while sitting in a bathtub, seeing how his own body increases the level of the water. In the background of the story, there was a craftsman, who was tasked with delivering a votive crown for King Hiero II of Syracuse. The king asked Archimedes to determine if the craftsman used all the silver he had obtained from the king - or maybe he had hidden some in his own pocket?
Using the newly discovered law of floating, Archimedes could calculate the density of the crown, a challenging task considering the sophisticated shape of a jewel. And determine if - in the end - the craftsman had committed fraud.
What industries and types of digital products are especially vulnerable and prone to fraud?
According to the recent PwC (former Pricewaterhouse Coopers) survey, nearly a half (47% to be precise) companies have experienced a fraud during the past 24 months, and it was everything but a single incident. On average, companies reported experiencing 6 incidents of fraud in the last 24 months.
Among the three most common types of frauds there were:
Customer fraud - experienced by 26% of companies, this form is the most common and especially seen in Financial Services and Consumer Market sectors. It is not only the most popular but also the most disruptive. The bad news is that an increasing number of companies shift toward the direct-to-consumer model. The good news is that this type of fraud is rarely sophisticated and the basic defense that takes the form of procedures, automation, and technical support is usually enough.
Third-party companies - outsourcing is getting an increasingly popular way to optimize costs and boost operational capacity. But the business partner can also be a source of fraud - and that’s a situation experienced by 19% of companies. It is a relatively new situation, where about half of companies surveyed lack the risk program and 21% have no due diligence program regarding their business partners.
Senior management - probably the most vicious source of fraud and relatively common (experienced by 26% of companies). The senior management team has both - the power, the knowledge, and the influence - to commit a devastating fraud rarely noticed.
What are the main threats?
First and foremost - fraud means losing money. According to the cited PwC study, there were up to $42 billion losses in the last 24 months. But money is only the tip of an iceberg when it comes to devastation delivered by fraud. Being the victim of fraud reduces the company’s competitive advantage and can be disastrous from the customers’ point of view. Companies that are frequent targets and victims of fraud are frequently noticed by media vendors - not in the way they would appreciate.
Fraud is also undermining employee morale and probably results in losing future business opportunities. That applies both to internally committed frauds like insider trading or money launderings, and external ones, like bribery or corruption.
The PwC states that 6 in 10 organizations don’t have any program to address bribery and corruption risk. Also, nearly half of respondents don’t perform a risk assessment or due diligence or ongoing monitoring of the third-party services providers.
Considering the size of the threat and the potential to disrupt the company’s operations, it is not a surprise that there is an increasing interest in using the AI-powered solutions to combat the fraud.
How can AI support fraud detection?
Financial services have a bit of an unfair advantage when it comes to applying AI in the solutions. These companies are heavy data-users with a long history of gathering and analysis did not only to make the business run smooth but also to keep compliance with the legal norms. Considering that, even the messiest and unorganized company needs to keep an eye on fiscal documents to avoid fines, and companies that have their core business in the financial sector need to keep their records clean, organized, and legible. For fintech, fraud detection is the competence that they simply need to ace.
Thankfully, the market is full of machine learning-powered solutions to leverage. Below we have collected only the most important or intersting ones:
The company was established to tackle the claim automation in the insurance company. They claim abuse is one of the most popular types of fraud in the insurance company, where customers appear to unleash nearly unlimited creativity to get just-a-bit-more than they should according to the company’s policy. Shift technology aims both to empower insurers to tackle the fraud while keeping to deliver customer excellence.
The not-the-top-of-mind example of fraud prevention - Red Points aims to protect the brands and their content in the digital world. The company scans the internet in search of violations of copyright, so it basically tackles a huge problem of counterfeits or illegal copies. And not everyone realizes that copyright infringement is also a type of fraud.
While most of the companies that aim to tackle fraud take their roots from the fintech business, buguroo hails from the cybersecurity background, initially offering the AI-powered platform that aims to defend the company against hackers. After a pivot the startup decided to shift from a typical cybercrime and DDoS attacks to any type of fraudulent activity that follows patterns, in the same way, the more typical hacking activities do.
Delivering reliable authentication methods and identity management that leaves no place for a malicious outsider to sneak in is powerful fraud protection. And that’s basically what Authenteq delivers - the blockchain and AI-powered identity verification tool. It can be compared to a biometric virtual ID card.
The company offers an Artificial Intelligence-based software platform that takes data from various sources and analyses the transaction to spot the pattern of fraud. It is possible due to the real-time monitoring of several thousands of data points and building an overall image of the user and his or her trustworthiness.
Calltic delivers the tech to detect a telecom fraud. Again, this type gains not that much attention like financial sector fraud, but it can be devastating for the customer. The most basic fraud scheme is on cybercriminals acquiring a premium number and generating high-volume international traffic on it using illegal or abusive techniques.
The company can be compared to a virtual auditor, that spots errors in financial data - in fact, a flagship product of the company is called AI Auditor. The advantage is that the analysis is done lightning-fast on the whole database, so tests can be run and re-run frequently.
The company blends the fraud protection and cybersecurity to provide the unified platform delivering early-warnings against hacking attempts or fraud attacks. The company aggregates data from a myriad of sources and detects attacks run from both surface web and deep web.
As mentioned above, identity management is crucial to building a fraud-hardened organization. Considering that, an account-hijack is a great threat for nearly every organization. An account considered trustworthy could have been taken over and used for malicious deeds - for example to conduct multiple fraudulent transactions or withdraw money from the owner. The technology delivered by SpyCloud actively scans the system in search of account-hijacking.
The company uses social and openly available data powered by Ai-based solutions to spot fraudulent transactions in real-time. The system is used mostly in online gambling, e-commerce, and streaming services.
Fraud can be seen as a hydra with many heads that regrow when cut. But the mythological hydra was not invincible. Hercules, a super-human hero of ancient myths, managed to outsmart the monsters regenerative abilities and by cleansing the wounds with fire he prevented heads from regrowing.
So is that with the modern fraud - a pure, brutal force of manual checking is not enough, as the hackers and criminals deliver increasingly sophisticated schemes of attacks. But when supported by artificial minds, companies are far from being defenseless.
Would you like to know more about fraud-detection technology? Are you interested in the advantages it can deliver for your company? Don’t hesitate to contact us now!