In the second of a two-part article series, Jonathan D Hancock, head of product and innovation at The AI Corporation, looks at new and emerging types of fraud, and how fleet managers can use AI to help them beat fraud in the future.

Jonathan D Hancock, head of product and innovation at The AI Corporation

In my previous article, I examined the impact of generative AI on fleet and mobility payment fraud. In this piece, we will continue that analysis to determine if generative AI has had the anticipated effect. While the technology has become readily available, the question remains: have fraudsters adapted their methods? Early indications suggest they already leverage generative AI to refine their attacks, creating more targeted, realistic, and sophisticated fraud schemes.

AI-generated phishing scams

One of the more immediate threats is using generative AI to produce more convincing phishing emails, messages and phone calls. Using NLG (natural language generation) capabilities, AI can craft personalised messages that are nigh impossible for recipients to distinguish from legitimate communications. Fraudsters could quickly generate fake notifications that appear to come from trusted sources, such as fleet management companies or payment processors, tricking employees into providing login credentials or erroneously authorising fraudulent transactions.

AI in social engineering attacks

Generative AI can also enhance social engineering techniques. Fraudsters could use AI-driven deepfake technology to impersonate company executives or fleet managers in video or audio calls, gaining access to confidential information or instructing employees to make fraudulent payments. With voice synthesis and video manipulation, the potential for such scams to succeed has increased dramatically.

Synthetic identity fraud

Generative AI’s capacity to create realistic and detailed personal data means fraudsters can use it to craft entirely synthetic identities. These synthetic identities, which mix real and fabricated data, can be used to apply for fleet cards or manipulate payment systems to gain unauthorised access to funds or resources. Since these identities could mimic real people so convincingly, spotting this type of fraud becomes much more difficult without advanced deep learning AI-driven detection mechanisms.

Can generative AI help us to beat fleet card fraud?

Fleet card fraud remains one of the most prevalent forms of fraud within the fleet and mobility industry. Fraudsters often target fleet cards due to their widespread use and relative lack of stringent security measures. But could generative AI finally turn the tide in the battle against fleet card fraud?

Enhanced fraud detection and prevention

Generative AI is already playing a significant role in improving the detection of fleet card fraud. By analysing transaction patterns, generative AI can identify subtle anomalies in card usage that might go unnoticed by traditional systems. AI can also simulate and predict fraud patterns, helping security teams stay one step ahead of fraudsters. Additionally, AI-driven predictive models can help prevent fraud before it occurs by flagging transactions that deviate from standard usage patterns. For example, AI can trigger alerts for further investigation if a fleet card is used in an unusual geographic location or for an unusually high amount of fuel versus known vehicle odometer readings.

AI-driven authorisation systems

Another promising application is the use of AI in the card authorisation process. Fleet card processors are beginning to use AI models to perform real-time checks on transaction legitimacy based on various factors, including user behaviour, geolocation and even the vehicle’s telemetry data. If something seems unusual, such as a fleet card used in a non-designated region or for an unauthorised purchase, the AI can automatically send a message to the fleet authorisation platform and decline the transaction.

Additionally, AI is improving the security of fleet card authentication. For example, biometric verification (fingerprint or facial recognition) is becoming a key part of fleet card security, ensuring that only authorised personnel can access the card or make payments.

Can generative AI help to beat forecourt security?

While many discussions around AI and fraud focus on online and e-commerce fraud, the physical world remains a crucial battleground. Forecourts, petrol stations and charging points where fleet vehicles refuel represent a vulnerable link in the chain. Fraudsters could use generative AI to manipulate systems at the point of sale or to circumvent fuel payment systems.

Automated fuel skimming

One potential threat is AI-driven fuel skimming, where fraudsters use generative AI to manipulate or mimic legitimate fuel point of sale (POS) terminals. AI could enable attackers to create counterfeit card readers or payment kiosks that look identical to legitimate ones, tricking drivers into entering sensitive card information. Once entered, the fraudster could use generative AI to decode or bypass the security features of the fleet card systems.

AI-enhanced skimming devices

Generative AI could also improve the design of physical skimming devices. These devices, which capture payment details at the point of sale, could be made more sophisticated by using generative AI to mimic the precise functioning of legitimate systems, making detection by traditional security methods much more difficult.

AI-assisted manipulation of telematics data

AI could also be employed to manipulate telematics data, allowing fraudsters to make it appear as though a vehicle was refuelled at a legitimate forecourt when, in fact, somebody stole the fuel. If fraudsters can bypass telematics systems that track fuel usage and vehicle data, this could result in significant financial losses.

A double-edged sword

Generative AI will continue to make strides in the fight against fleet and mobility payment fraud. Its ability to simulate, detect anomalies and predict future fraud patterns has already made a notable impact. However, as with any powerful tool, it comes with its own set of risks.

While businesses can use AI to stay ahead of fraudsters, those fraudsters can also leverage generative AI to refine and develop new, more sophisticated exploitation methods.

The retail fuel, fleet and mobility payments industry must continue to evolve and invest in its fraud prevention strategies by adopting cutting-edge AI-driven systems while ensuring that human oversight and comprehensive security measures are in place.

AI does not provide a one-size-fits-all solution, but with proper integration and layering into existing fraud management and processing platform frameworks, it can significantly improve detection rates and reduce the overall impact of fraud.

Ultimately, the success of generative AI in fleet fraud management will depend on how quickly both defenders and attackers adapt. As AI technology advances, fraudsters will likely continue to innovate and evolve, but with the right defences in place, fleet and mobility payment and fraud management systems can stay one step ahead.