Beyond the billion-dollar banking oversight: How process intelligence can surface vital warning signs

When one of Australia's Big Four financial institutions recently self-reported over $1 billion in potentially fraudulent loans, the industry's focus immediately turned to the sophistication of the bad actors. But for those of us looking at the mechanics of global banking, the more pressing question isn't how the documents were doctored, it's how the process allowed them to move through the system undetected for so long.

This isn't just an isolated compliance failure; it's a symptom of "process debt" - the accumulation of manual workarounds and outdated shortcuts - in mortgage origination.

This case raises a fundamental question for every financial leader: If deviations of this magnitude can occur, where else is the gap between how a process is designed and how it actually runs?

In complex environments such as mortgage origination, fraud often hides within the small workarounds, skipped steps, and manual overrides that become standard practice over time.

Surfacing the ghost paths

In high-volume lending, risk often accumulates in the shadows of non-standard workflows. Process Intelligence enables institutions to monitor the entire loan-origination lifecycle in real-time, highlighting where applications might be bypassing standard verification steps.

Rather than speculating on the specifics of any one case, we look at the systemic patterns that typically precede these events:

• Bypassed Controls: Identifying ghost paths where mortgage applications consistently skip mandatory income verification or fail to trigger standard document-check rules.
• Referrer Anomalies: Monitoring high-risk process paths to see if specific broker or referrer channels are consistently short-circuiting normal escalation steps.
• Structural Deviations: Detecting patterns in shell-company lending, such as multiple applications tied to entities that appear to meet only the bare minimum requirements for trading history.

From reactive audit to continuous monitoring

Traditionally, banks rely on retrospective audits - looking back after the damage is done. However, when AI is used to generate fraudulent documents at scale, the speed of the "bad" process often outpaces traditional human review.

By turning process-level data into real-time risk indicators, institutions can move towards a model of continuous process oversight. This doesn't just flag a single suspicious document; it surfaces clusters of risk. For example, identifying groups of borrowers, brokers, and entities that repeatedly appear together across siloed systems can reveal coordinated networks that would otherwise remain invisible.

Closing the execution gap

Fraud thrives where there is a gap between policy design and real-world execution. If a referrer program is designed with strict compliance but executed with manual workarounds to meet volume targets, the system becomes vulnerable.

Process intelligence helps financial institutions close this gap. By correlating data across different systems, it can identify deposit-source anomalies such as overseas deposits that deviate significantly from standard norms and flag them for investigation before the loan is finalised.

The ongoing investigations serve as a reminder that what we don't see can indeed hurt us. As fraud becomes more sophisticated through the use of AI, the defence must become equally sophisticated. The goal is no longer just to find the needle in the haystack, but to make the entire haystack transparent.

The technology exists today to ensure these warning signs move from hidden in the data to actionable intelligence. In a world of automated fraud, transparency isn't just an operational goal, it's a systemic necessity.

To learn more, visit celonis.com/solutions/banking.