The Hidden Risks of AI-Generated Code in Banking Systems

The rise of generative AI in software development promises to revolutionize productivity, but for the banking sector, this shift comes with hidden risks, as stated in FinTech Futures. While AI can accelerate coding processes, experts warn that over-reliance on AI-generated code without deep technical oversight could lead to inefficiencies, security vulnerabilities, and a looming skills crisis.

Dharmesh Mistry, a seasoned programmer with decades of experience in banking technology, draws parallels between today’s AI-driven coding and past shifts in programming paradigms.

«I wrote my first code at 12 on a mainframe, and by 17, I was a junior programmer at Lloyds Bank,»Mistry recalls. «Back then, transitioning from low-level languages like C to high-level tools like Visual Basic (VB) was a productivity game-changer—but it came at a cost.» While VB made Windows app development accessible, Mistry observed that programmers who relied solely on it struggled with complex system-level challenges.

«Programmers working exclusively in VB couldn’t solve intricate problems because they lacked fundamental knowledge of how the Windows OS worked. The best solutions still came from experienced C developers who understood the underlying system.»

AI and the Future of Banking Software

Today, large language models (LLMs) offer a similar promise: faster, more efficient coding. However, Mistry warns that banks must not overlook the need for «system programmers»—developers with deep architectural expertise who can supervise AI-generated code.

«AI will make programmers more productive, but we still need those who truly understand the system,» he says. «They must ensure AI-generated solutions are performant, secure, and resource-efficient—not just functionally correct.» Without this expertise, banks risk accumulating technical debt—much like the current shortage of COBOL programmers maintaining legacy systems.

The Urgent Need for Investment in Deep Technical Skills

Mistry emphasizes that banks must proactively train the next generation of system programmers rather than assuming AI will replace the need for expertise.

«It will be tempting for companies to cut costs by relying solely on AI, but that’s a dangerous gamble. Banks that maintain a core of deeply skilled programmers will avoid the crisis awaiting those who neglect expertise in favor of automation.» His warning is clear: AI is a tool, not a replacement for knowledge. The financial institutions that invest in system-level programming skills today will be the ones best positioned to harness AI’s potential without falling into technical insolvency tomorrow.

Other articles
Listening Is the New Power Move in Financial Services
SymphonyAI Eyes Agentic Automation as the Future of AML Compliance
AI Adoption in Financial Services and Fintech in 2025: Key Trends and Use Cases
Visa Launches New Initiative to Simplify Embedded Payments for Businesses
JPMorgan Unveils AI-Powered Tool to Combat Payment Fraud in Corporate Transactions
New ‘Buy Now, Pay Later’ Rules to Benefit Big Lenders, Not Hinder Them
Cable Insurance and TruckerCloud Join Forces to Improve Commercial Auto Data Systems
In-Car Payments Becoming Must-Have Feature for Drivers, Study Finds
Digital Wallets Are Evolving — And They Want to Replace Your Apps, Not Just Your Cards
Parents Call for Financial Education as the New “Fourth R” in Schools
The Role of AI-Driven Large Transaction Models in Transforming Payment Security
How Generative AI Is Fueling the Future of Embedded Finance
How Amazon and Walmart Are Shaping Retail’s Future With Robotics and AI
ECB Collaborates with FinTechs and Banks to Shape the Future of Digital Payments
The Top 10 Automotive Industry Trends to Watch (2025–2027)