Cybersecurity and Trust: why AI delivers for financial institutions

Financial services institutions are built on trust–sometimes over the span of centuries. In today’s world where everything is digital, cybersecurity plays a very significant role in building and maintaining that trust.

As it has been from the beginning, cybersecurity is an arms race: emerging technology does not discriminate between good and evil; the bad guys have the same tools as the good guys and each side is always trying to stay ahead of the other.

It is impossible to discuss emerging tech without discussing the rapid rise of artificial intelligence, which continuously demonstrates its real-life applicability. Large Language Models (LLMs) are all but commoditised and easily accessible to the general public and AI has certainly risen above “buzzword status”.

Emerging tech is tricky for financial services institutions who must take a risk-averse approach to build on the trust they have earned for their customers, but their criminal adversaries have no such limits. Now is the time for financial players to understand and embrace what AI can do for them on a security front and fight fire with fire.

When AI is seen as a hammer, everything starts to look like a nail. Despite the rise of AI and other emerging tech, effective cyber defense is still predicated on tried-and-true methods of prevention and detection; attackers–even artificial ones–have patterns.

What is different today is the breadth of the landscape and the volume of data. Non-human threat actors will likely be more precise and leave a much smaller footprint in comparison to their human counterparts, which makes these patterns more difficult to detect by human eyes.

AI really shines in its ability to analyse vast amounts of data quickly and accurately and finding patterns in the smallest crevasses of data which would take humans much longer to spot, if at all. What’s more is that effectively implemented AI systems can learn over time, adapt to new threats, and even become proactive in predicting attacks before they happen based on indicators of compromise, making it one of the most powerful tools of modern cybersecurity.

The output of AI offers a compelling proposition for risk-averse companies as it can quantify and manage risks in ways that humans can’t. When AI is provided historical data, it can be trained to forecast potential security issues and even propose ways to mitigate them before they come to pass, informing a potential focus of resource allocation to avoid costly incidents.

Considering AI’s pattern recognition capability, it has other applicability in the security domain beyond cyber threats. For example, AI-powered biometric systems and user behaviour analysis can detect indicators of fraud with less false positives than ever before. AI can also be trained to maintain regulatory compliance by analysing transactions for suspicious activity and raising anything which may require more judgement to help ward off financial crime and reputational damage.

The benefits of AI are many, but bringing this technology into a financial institution is not without challenges. In many cases, AI is an opaque system which does not offer transparency around its decision-making, which is a high contrast to the needs of these institutions, making it problematic for accountability and compliance purposes.

Data privacy is also a concern, as an AI system requires access to a vast amount of data which has to be considered. Each institution will have to decide for themselves where the balance between security and privacy lies. Also, whilst AI is great at analysis, human judgement is often still required, which means AI-powered tools must be refined and staff must be trained to make the most of them.

Without a doubt, AI is here to stay. It is a real technology solving real problems, and it provides the most value when it is finely scoped and tuned. AI, for now, is not a replacement for cybersecurity professionals.

In early 2024, the United States government flew actual military fighter jets in a dog-fight scenario, where one jet was piloted by a very experienced aviator, and the other piloted by an artificial intelligence trained in dog-fighting, but the AI-driven jet still had a human pilot in the cockpit to take over if needed. It is worth noting that the pilot reportedly did not take control of the aircraft at any point during the scenario, and they did not disclose the victor.

Financial institutions should be taking a hard look at how to add AI to their security toolbox to help their already skilled and trained stewards identify threats quicker and more precisely, as it is not an all-or-nothing implementation. AI, despite some of its risks, can be implemented safely and its risks mitigated if done methodically using the tried-and-true methods upon which these institutions have built their foundations.

---

Brian Wagner is a seasoned expert in cloud technology and security with over 20 years of experience, particularly in the global financial services sector. He has held senior leadership roles, including Head of Security and Compliance at AWS Financial Services EMEA, Director of Cloud and Security at a Silicon Valley analytics firm, and CTO of a UK-based cybersecurity provider. As CTO at Revenir, Brian leverages his expertise to develop innovative products tailored to the financial services industry’s unique needs, ensuring security, scalability, and reliability.

• Read more under:
• Artificial Intelligence
• BankingTech