Smarter than the scam: how optimized AI is reshaping fraud detection
Table Of Content
Fraud remains a persistent threat for businesses of all sizes, with the trade association UK Finance estimating that over £1.1bn was lost to fraud in the UK alone in 2024. Financial services firms are a particularly attractive target, given the opportunity for significant monetary gain.
To put this into context, Q1 of 2024 saw 8,374 consumers lodge complaints about fraud and scams, of which over half were in relation to customer approved online bank transfers. These types of scams that often involve convincing social engineering and AI-generated content, which manipulate people into sharing confidential information, are becoming harder to spot- even for experienced professionals.
For financial services providers, the consequences of fraud reach far beyond lost revenue. Companies can suffer reputational damage, operational disruption and costly legal fallout, all of which can undermine long-term business resilience. Unsurprisingly, these firms take a proactive approach to fraud prevention.
But with fraudsters becoming more sophisticated and technologically advanced, especially with the help of AI, it can often feel like a high-stakes game of cat and mouse, with defenders constantly reacting to ever-evolving tactics.
Director of Analytics and Automation at Sopra Steria.
Fraud is not disappearing anytime soon
The National Crime Agency ranks fraud as the most common crime in the UK, accounting for 41% of all crime in England and Wales. For financial services firms, the challenge is even more acute. In 2024 there were 3.13 million confirmed cases of fraud, a 14% rise compared to 2023. This sharp increase highlights a two-fold problem, whereby fraudsters are becoming more sophisticated, whilst the overall volume of attacks is climbing.
There are various reasons behind the rising levels of fraud, not least the ongoing cost of living crisis. As operating costs soar and cashflow becomes tighter, financially vulnerable businesses are more likely to take risks, overlook red flags, or fall for offers that seem too good to be true.
This makes them an easier target for scammers. On top of this, AI is transforming how fraud is carried out. Specifically, it enables criminals to move faster and more easily deploy advanced tactics, from deepfakes and synthetic identities, to highly convincing phishing campaigns.
Optimizing automation is the remedy for success
One of the main reasons financial services providers remain prime targets for fraudsters is the sheer number of customer interactions across mobile apps and online banking platforms, creating multiple potential entry points that are difficult to monitor and secure simultaneously.
This challenge is often compounded by a lack of alignment between key teams, such as fraud prevention, customer authentication and customer service, leading to gaps in visibility across the customer journey. With high transaction volumes, countless access points, and siloed teams, it can be difficult for firms to keep false positives to a minimum and contextualize and action suspicious activity or alerts.
Financial services companies often turn to AI to help reduce false positives which, in principle, is the correct way to go. However, many AI tools still rely heavily on manual rule creation and editing, leaving fraud teams to handle the hardest tasks, such as deciding what patterns to target and making sense of ambiguous data.
Instead, firms must go beyond rule-tweaking and help uncover the ‘why’ behind fraud, not just the ‘what’. To get the best value from AI, financial services providers should turn their attention to models that can enhance the decision engine, by replacing legacy rulesets in favor of a new auto-generated suite of optimized rulesets.
In practice, this approach starts by analyzing historical transaction data and fraud model scores to identify patterns in both legitimate and fraudulent activity. It then determines the right balance between detecting fraud and avoiding false positives.
From there, a fresh set of optimized rules is automatically generated and can be refreshed daily. This keeps the fraud decision engine sharp and responsive, reducing manual effort while ensuring it stays aligned with the latest threat patterns.
The benefits must be balanced with human input
Optimizing fraud detection leads to a more efficient and accurate program, enabling fraud teams to focus their attention on genuine scam activity without being distracted by false positives. More broadly, it allows firms to respond quickly and effectively to shifts in risk appetite, tightening or relaxing controls as circumstances change.
While these benefits of AI and automation are undeniable, financial services firms must not become over reliant on these technologies. Human expertise remains essential, especially in scenarios that demand emotional intelligence, nuanced behavioral analysis or complex judgement.
AI is a powerful tool, but it’s the insight and oversight of skilled fraud professionals that ensures investigations are handled with the necessary sensitivity and rigor. Crucially, it’s also down to fraud teams and data scientists to train and fine-tune AI models using real-world insights, ensuring they evolve in line with the shifting threat landscape.
Plan for long-term fraud resilience
Blending human expertise with AI isn’t just a technical enhancement, it’s critical for any financial services provider that is serious about safeguarding customers and staying ahead of increasingly sophisticated fraud. As threats evolve, particularly with the rise of agentic AI, relying solely on reactive measures won’t be enough. Proactive, adaptive fraud prevention, which combines the precision of AI with the intuition of human judgement, will be key to preserving trust and resilience in the long term.
We’ve featured the best online cybersecurity course.
This article was produced as part of TechRadarPro’s Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro
No Comment! Be the first one.