There is no shortage of activity across financial crime functions. Systems are deployed, alerts are generated, teams are expanded, and reporting continues to scale.

The question is whether any of that is translating into a meaningful reduction in harm.

In a recent Transform Finance podcast, Ray Blake, FinCrime subject matter expert and Director of Dark Money Files, challenges several assumptions that continue to shape how institutions approach financial crime. The picture that emerges is less about capability gaps, and more about structural misalignment.

The Software Illusion

There remains a persistent belief that financial crime risk can be materially reduced through the procurement of better technology.

Not improved. Reduced.

The distinction matters. The idea that a system can be implemented and relied upon to neutralise exposure with minimal ongoing intervention continues to surface across the industry.

It does not hold.

Technology can support decision-making. It does not replace it. More importantly, it does not absorb accountability. Regulators do not hold systems responsible for failures in financial crime controls. They hold firms, and ultimately individuals.

The risk is not the use of technology. It is the expectation placed on it.

Measuring What We Do, Not What Happens

Much of the industry’s reporting still centres on activity.

SAR volumes, alert numbers, and enforcement outcomes remain the dominant indicators of performance. These are easy to quantify, easy to benchmark, and easy to present.

They are also detached from the core question.

Is financial crime being reduced?

There is limited evidence that these metrics provide a reliable proxy for harm reduction. Regulatory fines, often cited as evidence of deterrence, are particularly weak in this regard. At a global level, they represent a negligible fraction of financial system profits.

If the cost of failure is absorbed as a cost of doing business, the signal becomes difficult to interpret.

AI and the Structure Problem

The current AI narrative suggests a step-change in detection capability. In practice, much of its deployment reflects a continuation of existing approaches.

The prevailing model remains intact:

• Transaction monitoring systems generate large volumes of alerts
• A small proportion is escalated
• The majority is dismissed

AI is increasingly being applied to manage this process more efficiently.

What it is not doing, in most cases, is addressing why the volume exists in the first place.

If the upstream system continues to produce excessive noise, downstream optimisation does not resolve the underlying issue. It compresses the cost of managing it.

This is not a marginal observation. It speaks to where investment is being directed, and what is being left unchanged.

Where Controls Actually Succeed or Fail

Structural debates, including the separation or integration of fraud and AML functions, remain active across institutions.

They are not insignificant. They are also not decisive.

The point at which financial crime is most effectively identified is earlier, and more exposed.

It sits at the customer interface.

The ability to assess whether activity makes sense, in context, in real time, remains one of the most effective controls available. When that judgement is weak, no amount of downstream processing compensates for it.

Reorganising teams does not address this. Strengthening judgement at the point of interaction does.

Regulation Without a Baseline

There is a tendency to assess regulatory effectiveness through visible activity. More enforcement, higher fines, increased scrutiny.

The interpretation is less straightforward.

Increased enforcement can signal stronger oversight. It can also indicate that underlying issues remain unresolved. Without a credible baseline for the scale of financial crime, neither interpretation can be confidently validated.

Estimates of global laundering volumes continue to circulate. Their precision remains questionable.

The result is an environment where activity is visible, but impact is harder to evidence.

The Risk of Looking Backwards

Control frameworks are often shaped by what has already happened.

This is understandable, but also limiting.

There is a tendency to embed responses to historical typologies into long-term control structures. Over time, these controls become part of the operating environment, regardless of whether the underlying threat persists.

The effect is subtle. Effort is concentrated on known patterns, while emerging risks receive less attention.

Financial crime does not operate on that timeline. It adapts quickly, and often outside the areas receiving the most scrutiny.

What Is Underestimated

Looking ahead, the gap is not purely technological. It is anticipatory.

Financial crime professionals operate with a detailed understanding of systems, controls, and vulnerabilities. That perspective is rarely used to model how those systems might be exploited next.

More time is spent analysing what has happened than stress-testing what could happen.

That imbalance creates exposure.

Operational Friction

There is a final constraint that cuts across all of this.

Financial crime is executed with speed and coordination. Institutional responses are often distributed, segmented, and dependent on multiple handovers.

Each handover introduces delay.

Each delay reduces effectiveness.

Improving outcomes, in this context, is not only about better detection. It is about reducing friction across the response.

That is an operational challenge as much as a strategic one.

Closing Observation

Financial crime functions are not underactive. If anything, they are saturated with activity.

The issue is whether that activity is directed at the points of highest impact.

Until measurement, structure, and strategy align more closely with outcomes, the industry risks continuing to optimise effort without materially shifting exposure.

Watch the Full Podcast Here!