There are many cases of professional use more or less convincing artificial intelligence (AI). But there are two that interest more and more organizations: detecting and deterring fraud. This is demonstrated in a study conducted in partnership with the Association of Certified Fraud Examiners (ACFE) and the SAS analysis company.
The study, entitled "Anti-Fraud Technology Benchmarking Report", is based on responses from more than 1,000 CAFE members from all sectors and around the world. The goal was to find out how they used AI to reduce fraud.
One of the most interesting findings of the study is that only 13% currently use AI for this purpose. But a quarter plan to start doing this within two years.
Most recent threats
This is not surprising considering the benefits that artificial intelligence could bring to the detection of fraud. For example, the AI can determine if an interaction with a company does not match typical transaction characteristics.
To do this, you have to look at many features in seconds, which no expert can succeed. Some AI solutions can also detect various types of fraud without prior exposure. Such an advantage allows business systems to stay up-to-date and face the latest threats.
The study also showed that half of the companies plan to spend more of their budget on anti-fraud technology over the next two years. Almost one in four organizations already uses biometrics to stem fraud. 16% plan to use it by 2021.
Companies also plan to adapt the way they analyze data. Automation should be one of the most successful solutions: a vast majority (72%) of organizations plan to use automated monitoring, anomaly detection and exception reporting by 2021. This strategy will help probably save time and avoid false alarms.
Reduce the number of false positives
The results of this survey also indicate that 52% of respondents plan to rely on modeling and predictive analysis (an increase of 22%). Predictive analytics could help them determine which types of threats are most likely to impact their business.
Banks and financial institutions could benefit from a fraud detection solution based on machine learning or machine learning.
According to a case study presented by Teradata, the AI has allowed Danske Bank (the largest bank in Denmark, which is also at the heart of a scandal based on 200 billion euros of suspicious transactions ...) to modernize its Fraud detection process and reduce the number of suspected false positives of 1,200 per day.
The AI solution that has been selected can evaluate credit card transactions, online and via mobile in less than 300 milliseconds. For its part, Mastercard relies on the detection of fraud via the AI to reduce the number of fraudulent transactions and the number of times customers refuse transactions when everything is fine. Mastercard's technology can reduce the rate of transactions that have been inadvertently declined by 50%.
By combining supervised learning algorithms built on historical data with unsupervised learning, companies could gain more insight and clarity about the relative risk of client behavior.
With an artificial intelligence-based fraud prevention system - which evaluates historical data and anomalies - the customer experience is not impacted.
