The Hidden Cost of Frictionless Oversight

Over the past several months, we've explored a recurring theme: compliance leaders are operating in an environment defined by increasing velocity.

Business is moving faster. Firms are expanding into new products, new technologies, and new markets. Regulatory frameworks are becoming less prescriptive, but the laws haven't changed and accountability remains firmly in place. Increasingly, firms are expected not only to have controls, but to demonstrate how those controls worked when examined in retrospect.

The findings from the 2026 StarCompliance Global Compliance Benchmark Study suggest these pressures are not anecdotal. Compliance budgets are increasing. AI adoption is accelerating. Surveillance programs are expanding. Firms are investing heavily to keep pace with growing complexity.

Yet one finding stood out.

Despite significant investment in compliance technology and AI, most respondents reported little or no meaningful reduction in workload. In many cases, technology reduced manual effort in one area while creating new responsibilities around governance, oversight, and review in others.

A Technology Problem? Or a Systems Problem?

That finding deserves more attention than it is likely to receive.

The natural explanation is that the technology is still evolving. Given enough time, better models, better automation, and better tools will eventually deliver the efficiency gains firms expect.

Perhaps.

But the benchmark data points toward another possibility.

The industry increasingly talks about AI as the challenge. The study suggests many firms may be struggling less with technology itself and more with the operating models surrounding it. Across the survey, familiar themes appear repeatedly:

• fragmented systems
• disconnected workflows
• data silos
• governance complexity, and
• growing oversight burdens

In other words, many organizations are introducing powerful new technology into environments that were designed for a different era.

That approach can certainly make existing processes faster. Whether it improves decision-making is a different question.

Why Friction Is Good

This distinction matters because compliance is not primarily an efficiency function. It is a judgment function.

The decisions that matter most rarely involve straightforward rule application. They involve context, ambiguity, competing priorities, and incomplete information. Whether reviewing an employee trading exception, assessing a potential MNPI exposure, evaluating an outside business activity, or determining whether a communication requires escalation, the challenge is rarely the process itself. The challenge is the judgment required to reach a defensible decision.

That is why the current focus on frictionless workflows deserves a closer look.

In most business functions, reducing friction sounds unquestionably positive. Workflows move faster. Bottlenecks disappear. Decisions are made more quickly.

Compliance is different.

Some forms of friction are not obstacles to good decisions. They are part of the process that produces them.

Challenge creates friction. Independent review creates friction. Escalation creates friction. Documentation creates friction. Even disagreement can create friction.

Yet those moments are often where assumptions are tested, weak reasoning is exposed, and important risks surface before they become larger problems.

The Design Challenge Ahead

This becomes particularly relevant as firms integrate AI into compliance workflows.

Many operating models follow a similar pattern. AI performs much of the analysis and a human reviews the output at the end, before a final decision is made. On the surface, this appears to preserve human oversight while increasing efficiency.

But there is an important difference between reviewing a conclusion and actively participating in the reasoning that produced it. At the risk of sounding annoying, I;m going to say that again but differently. Despite what the AI vendors say, having AI automate 4 out of 5 steps in a process, with the human at the end of the loop "rubber-stamping" the output is a very BAD way to design a compliance system.

Over time, people become increasingly less effective at challenging conclusions they did not help construct. The role gradually shifts from decision-maker to reviewer. Human involvement remains, but human judgment becomes less engaged and progressively less effective. Simple truth is the human cognition erodes in scenarios like this.

The risk is not that humans disappear from compliance. The risk is that they remain present while contributing less of the critical thinking organizations assume they are providing.

The benchmark study hints at this next phase of complexity. Compliance teams are increasingly responsible not only for surveillance and supervision, but also for model governance, workflow validation, AI oversight, data quality, escalation governance, and evidence traceability.

These are not simply technology challenges. They are systems design challenges.

As firms continue investing in AI, the most important question may not be what tasks can be automated. It may be where judgment should remain, where challenge should occur, and where a certain amount of friction should be deliberately preserved.

The firms that answer those questions well are likely to build stronger compliance programs than those focused solely on speed and automation.

As firms automate more compliance activity, are they improving judgment — or quietly eroding it?