William Wright
This article was produced in collaboration with Tech Trek Events for the NexGen Banking Summit 2025.
As criminals weaponise AI for sophisticated fraud attacks, banks need to improve fraud defenses. That’s why the NexGen Banking Summit 2025 will convene industry leaders in London this October to address this arms race head-on.
With senior leaders from institutions like JP Morgan Chase, the Bank of England, HSBC, and Citi taking the stage, the summit provides the platform for banks to understand and strategise routes to deploy GenAI.
This is the first article in a short series in the run-up to my attendance of the NexGen Banking Summit on the 15-16th October 2025. Check out the next article which will focus on the economic and political landscape of AI, considering how the UK stacks up to its peers in terms of balancing innovation and customer safety.
The AI Boom
As humans do with most major issues, the AI boom has been reduced to simple binaries: optimism vs pessimism, opportunity vs existential risk, good vs bad. On one hand, AI-powered innovation can be deployed across every industry to solve fundamental challenges and drive efficiency. On the other, the reordering of the job market could leave hundreds of millions displaced in a hitherto unimaginable mass technological unemployment.
In reality, the impact of AI is a complicated spectrum and one where most outcomes are not set in stone; each industry has a chance to define its future. Whilst this debate on AI’s long-term benefit rages on, one outcome is set in stone; bad actors will harness the power of AI to commit crime.
Real Crime, Right Now
Right now, criminals are leveraging different forms of AI to create hyper-realistic deepfakes, scalable social engineering attacks, and sophisticated, autonomous malware. Up from £2.2 trillion in 2015, global cybercrime is expected to cost £7.8 trillion this year; this will not slow down. By eliminating the need for technical expertise, AI shatters the barriers for entry that once restricted fraud to smaller, less efficient operations.
We are witnessing an ‘industrialisation of fraud’, with criminals easily able to automate fraud systems, rapidly scale their volume of attacks, and augment existing scams.
For banking, an industry plagued by legacy systems and slow implementation times, this risk is catastrophic. In a survey of over 400 senior fraud leaders and decision makers, 73% agree that GenAI has permanently altered the fraud landscape, making it more complex and sophisticated. And yet, less than 25% of banks have successfully integrated AI into their strategic playbook, with most AI projects failing to go beyond siloed pilots and proof of concepts. 56% of businesses struggle to identify the use of GenAI in fraud attacks. While most banks recognise AI-enabled crime as an urgent priority, few are dealing with it well.
The time for deliberation is over. Banks must open progressive dialogue and harness collective insights to overcome the shared challenges faced when implementing AI to fight fraud and meet regulations.
This is the leading imperative behind the 2025 NexGen Banking Summit. Delegates will gain access to real-world case studies, best practices from peers, and direct insights from the world’s leading fraud and risk experts. Sessions like “Generative AI & Ethics: Governance, Bias, and Responsible Innovation” and “AI, Cybersecurity, and the Evolving Threat Landscape” are designed to equip attendees with actionable strategies, offering a rare opportunity to learn from those on the front lines.
Understanding GenAI
AI is a dense and confusing term. AI is not one specific technology, but a broad range of technologies that allows computers to simulate some form of human intelligence. Take the first chess-playing computer in the 1910s, the first electronic voice synthesiser in the 1930s, or ChatGPT in the 2020s; these separate technologies can be grouped together as AI.
Generative AI, or GenAI, is a type of AI that learns patterns from large and complex data sets and can harness these patterns to create new and original content from a prompt.
A key aspect of most GenAI systems is that they have a ‘human-in-the-loop’, meaning human input is required to make these systems work. This is distinct from a ‘human-on-the-loop’ system where a human only provides oversight into an autonomous setup. A fully autonomous setup can be classified as an Agentic AI.
How can GenAI Improve Fraud and Risk Mitigation?
There are two main ways in which banks are deploying GenAI systems to adapt to the increased threat of AI-powered crime and improve risk mitigation.
Improved Reactive and Proactive Fraud Detection
When implemented properly, GenAI can both strengthen a bank’s reactive approach to detect and respond to fraud, and offer proactive solutions to anticipate and flag fraud risks before they escalate.
For a bank’s reactive approach, GenAI can monitor massive amounts of transaction, behaviour, and location data in real-time, using sophisticated learning models to instantaneously spot anomalies, assess risk with high accuracy, and generate synthetic data for model training to proactively prevent both known and novel fraud schemes.
This upgrades the traditional fraud systems using rule-based approaches and ill-defined risk signals such as spending static thresholds, transaction patterns, and location mismatches. As fraudsters get more sophisticated, they can easily bypass these systems and avoid triggering blocks and filters. By shifting to advanced GenAI systems, banks can move away from these static measures towards dynamic protections that safeguard customers.
In terms of a proactive approach to fraud detection, GenAI enables live anomaly detection through pattern recognition in real-time transaction data. GenAI also excels at generating synthetic data, or data not produced by real-world events. This allows for augmentation of datasets when there is a lack of high-quality data, enabling the training of robust detection algorithms.
When combined, these highly-trained, specialised algorithms learn about a bank customer’s patterns; how they spend? Where? How much? As spending patterns are dynamic, these algorithms will morph to better represent how a customer would spend. This means the complex frauds and scams being deployed can better be counteracted and more money saved.
Reinventing Regulatory Compliance
Evolved fraud detection with GenAI connects with another main focus area for banks: regulatory compliance. Banks are some of the most heavily regulated entities in the world but regulations vary by location. One only has to look at Revolut’s three-year struggle to gain a full banking license in the UK to know that fulfilling each and every expectation is difficult, time-consuming, and expensive.
Since the UK left the EU and passporting agreements were removed, financial institutions had to acquire licenses in both the EU and the UK separately if they wanted to carry out full operations. Considering most banks act internationally, this is a serious hurdle.
What makes this trickier is that regulations change all the time and continue to vary by location; when the UK makes a change distinct to Europe, this must be implemented across a bank’s entire UK operation – again, costly and time-consuming. Banks struggle to embrace dynamism and innovation due to the regulatory constraints.
GenAI and RegTech offer solutions, allowing banks to manage regulatory compliance processes more efficiently and cost-effectively. GenAI can automate various compliance processes while monitoring transactions in real time to mitigate risk and ensure real-time regulatory compliance.
Take onboarding, a tricky and regulation-heavy area for a bank. For Anti-Money Laundering (AML) and Counter-Terrorist Financing (CTF) checks, a bank must hold sufficient ‘Know Your Customer’ (KYC) or ‘Know Your Business’ (KYB) information. These are extremely expensive, time-consuming, and information-heavy checks to complete.
Banks can automate much of these processes by integrating GenAI into the KYC/KYB workflows. Currently, the central challenge in adopting these solutions is overcoming the limitations of public models (lack of proprietary data, no fact-checking) while ensuring that integrated AML/CTF systems remain transparent, auditable, and compliant with regulatory standards. By pooling resources and opening dialogues, banks can overcome these hurdles.
GenAI and the NexGen Banking Summit
The rise of AI-powered fraud is a significant issue, but also a test of how adaptive and collaborative the banking industry can be. The 2025 NexGen Banking Summit provides the necessary forum for this transformation. With demand already high and limited seats available, now is the best time to secure your place.
This article is the first in the NexGen Banking Summit Series. Stay tuned for the next piece which will examine the UK’s position in the global AI race, charting a path that not only encourages innovation but also builds a resilient and secure financial future.
Featured image via FreePik.
