Have you ever been locked out of an online account because the security measures were too strict? Or found yourself navigating endless hurdles just to prove you’re not a fraudster? You’re not alone. As fraud tactics evolve, so do the challenges of detecting them without creating unnecessary barriers for legitimate users.

A few weeks ago I attended the GovNet’s Counter Fraud conference. A gathering of the UK’s leading public sector counter fraud professionals that are exploring the ongoing challenge of balancing robust security with accessible, user-friendly services.

Fraud almost always starts with a lie

One of the event’s standout sessions featured Stephane Pendered from the Crown Prosecution Service and Zoe Gascoyne from HMRC, discussing the high-profile Bernie Ecclestone case.

What started as a media leak about bribery led to a civil investigation under HMRC’s COP9 process. Initially focused on unpaid taxes, the case escalated when investigators uncovered hidden trusts and financial misrepresentation, culminating in a £652M settlement.

The key takeaway was simple: fraud often begins with a small deception, whether for personal gain or at the taxpayer’s expense.

This lesson extends beyond high-profile cases. Effective fraud prevention isn’t just about sophisticated detection systems; it’s about understanding behaviour and recognising the early signs of dishonesty before they escalate.

The data challenge.

Many organisations face constraints such as legacy systems, restricted data access, and regulatory challenges. These limitations create a gap between what’s technically possible and what’s realistically achievable in fraud prevention.

Aman Virk from DWP emphasised the importance of trust and cross-team collaboration in overcoming these challenges. Her insights reinforced a key message: technology is only as effective as the human systems behind it.

The panel shared practical strategies to bridge this gap:

  • Creating safe spaces to test and learn quickly
  • Clearly defining value before investing in expensive technology
  • Securing leadership buy-in and aligning teams for successful scaling

Above all, they stressed that fraud solutions must not create unnecessary friction for legitimate users. The vast majority of people accessing public services do so honestly—security measures should reflect that.

AI in fraud detection: Potential and pitfalls

AI has been used in fraud prevention for years, but the focus is shifting from simply flagging more cases to identifying the right ones, cases that increase efficiency and add real value.

The event showcased several compelling applications:

  • Singapore’s government and law enforcement use AI to detect money laundering
  • The US IRS is leveraging AI to prioritise tax fraud cases and reduce refund fraud
  • Healthcare providers analysing unstructured data for fraud detection

However, two key questions emerged:

  • Does AI risk replacing fundamental fraud prevention measures?
  • How do we ensure AI increases efficiency rather than adding to workloads?

AI must be transparent and explainable to prevent unfairly flagging legitimate users. False positives aren’t just statistical errors, they create real frustration, delays, and barriers to essential services.

People first, technology second

The most important takeaway from the event wasn’t about advanced technology or detection systems.  It was about trust, alignment, and putting people at the heart of fraud prevention.

Effective fraud prevention isn’t just about stopping bad actors; it’s about ensuring that 99% of legitimate users can access services smoothly and securely. Strengthening data foundations, fostering collaboration, and designing security measures that enhance, rather than hinder, the user experience is key.