Leaving this course?

Lesson 1 of 5

Why the insider is a different problem

Almost every control an institution owns is designed to keep the wrong people out. An insider is already in, and three things follow that change the problem completely.

The access is legitimate. Nothing is bypassed, nothing is broken into. The system records authorised use by an authenticated user, because that is exactly what it is. No access control, however strong, distinguishes a clerk opening a customer record to serve that customer from the same clerk opening it to sell what is in it. Detection therefore cannot be about access; it has to be about the pattern of use.

They know the controls. Where thresholds sit, which reports are reviewed and by whom, when reconciliation runs, which alerts are noisy enough to be ignored, and who is on leave. An external attacker probes; an insider already knows. That is why insider schemes so often sit precisely below a threshold, or inside a window nobody reviews.

They have a reason to be looking. An unusual search by a stranger is an anomaly; the same search by someone whose job involves searching is a Tuesday. The contextual signal that gives away an outsider is absent.

What insiders actually do. Six shapes, roughly in order of how often they are found:

  • Theft of funds — unauthorised transfers, manipulated refunds and reversals, dormant accounts drained, loans created in false names.
  • Data theft and sale — customer records, card data, balances, identity documents, sold to fraudsters who then have the perfect basis for impersonation.
  • Procurement and vendor fraud — a supplier owned or favoured by an employee, inflated invoices, services not delivered.
  • Payroll and expenses — ghost employees, duplicated or inflated claims. Small, persistent, rarely prosecuted.
  • Information leakage to criminals — the subject of the next lesson, and the most damaging of the six.
  • Deliberate control weakening — alerts disabled, parameters widened, exceptions granted, logs disabled "for performance".

And the reason it is under-counted. Insider cases are frequently resolved as employment matters and never recorded as fraud at all, which — exactly as with first-party fraud — means nobody can see the size of the problem and therefore nobody resources it.

And "insider" is wider than "employee". The word suggests someone on the payroll, which leaves out most of the people who hold access:

  • contractors and temporary staff, often in exactly the high-access roles — systems administration, data migration, a project touching customer records;
  • outsourced service-centre staff, who may handle more customer records in a day than any branch and who are screened, supervised and disciplined by another company in another country;
  • cleaners, security and facilities staff, who have physical access to desks, printers and papers out of hours;
  • former staff whose access was never removed, which is a control failure rather than a category.

Each of these sits outside the institution's own screening, training and disciplinary processes while sitting inside its systems. An insider programme scoped to employees has scoped out part of its own problem.

Lesson 2 of 5

The insider as enabler

The most expensive insider is often not the one who steals. It is the one who makes someone else's crime possible, because the sums involved are not limited by what one person can carry out personally.

What this looks like in practice:

  • Opening accounts that should not exist — processing applications with documents known to be false, overriding verification, approving a customer the system declined. One cooperative member of staff can produce dozens of mule accounts that are, on the record, properly onboarded.
  • Selling customer data — balances, addresses, security answers, recent transactions. This is what turns a generic impersonation attempt into a convincing one, and it is why some scam calls contain details that only the bank could know.
  • Tipping off — telling a customer that their activity has been questioned, that a report has been made, or that an account is under review. A criminal offence in most jurisdictions, and devastating to an investigation. Worth being explicit: this is the point at which an insider case becomes an anti-money laundering case.
  • Bypassing sanctions or screening — clearing an alert that should have been escalated, amending a payment message, or re-entering a payment to avoid a filter.
  • Disabling or tuning out detection — raising a threshold, suppressing a scenario, closing alerts in bulk without review.
  • Providing operational intelligence — telling an outside party when reconciliation runs, when a manager is away, how a limit is approved.

Why it is hard to see. Each individual action is one an employee is entitled to take. The alert was cleared — clearing alerts is the job. The account was opened — opening accounts is the job. The record was viewed — viewing records is the job. The wrongness is in the relationship between that action and someone outside, which no system holds.

Where the signal usually is. In the correlation between a member of staff and an outcome: accounts they opened failing at a far higher rate than their peers'; alerts they cleared later turning out to be true; customers they served appearing in fraud cases disproportionately. This is peer comparison, and it is the single most productive insider detection technique available — it needs no surveillance of individuals, only the same outcome data the institution already holds, grouped by who handled it.

And the obligation. Where an employee is suspected of enabling money laundering or of tipping off, that is reportable in its own right, and handling it purely as a disciplinary matter leaves a legal obligation unmet.

Lesson 3 of 5

Collusion, and what it does to control design

Most control design assumes independence. Segregation of duties, maker-checker, dual authorisation and four-eyes review all rest on the same premise: that two people are not working together. Collusion removes that premise, and with it most of the control estate at a stroke.

The common pairings:

  • Two employees in the same process — a maker and a checker, an initiator and an approver. The oldest and most effective form, and it defeats maker-checker entirely.
  • Employee and customer — loans approved for a related party, valuations accepted that should not be, accounts operated jointly in fact though not on paper.
  • Employee and supplier — the procurement case, usually with an undisclosed interest or a kickback.
  • Employee and organised crime — the enabler relationship from the previous lesson, usually procured by payment, sometimes by pressure, debt or threat against family. Coerced insiders exist, and an institution that assumes every insider is simply greedy will mishandle them.
  • Manager and subordinate — the hardest to detect, because the authority that should challenge is the authority that is involved, and subordinates have little practical ability to object.

What this means for control design. Not that segregation is pointless — it raises the cost of fraud, requires conspiracy, and most people will not conspire. But two things follow:

  • Controls that depend on independence need a control that does not. Reconciliation performed by a third function, audit sampling, peer-comparison analytics, and data that cannot be altered by the people being measured.
  • The classic personnel controls are not bureaucracy. Mandatory consecutive leave, job rotation and enforced cover exist because many insider schemes require continuous maintenance — a reconciliation that must be fudged every month, an exception re-applied, a report intercepted. Taking the person away from the process for two weeks is one of the few controls collusion cannot easily defeat. Someone who never takes leave and insists on handling one task personally is a signal that costs nothing to look for.

The uncomfortable corollary about seniority. The more senior the insider, the weaker the control environment around them: fewer approvals above them, more ability to grant exceptions, more credibility when questioned, and a greater reluctance in others to escalate. Most control frameworks are strongest where the money is smallest.

The control that collusion cannot defeat is a colleague. It is worth naming explicitly, because it is left out of most control frameworks: a great deal of insider fraud comes to light because somebody noticed and said something — not because a system produced an alert.

Which makes the speak-up route a detection control, and it should be judged like one. Does a member of staff know where it is without looking? Can it be used without the line manager, who may be the problem? Is there a route that does not require certainty, so that "something about this feels wrong" is an acceptable report? Is anything heard back, ever?

A channel that exists but is used once a year across a few thousand staff is not working, and the reason is almost always that someone used it once and watched nothing happen.

Lesson 4 of 5

Detecting misuse of legitimate access

The question is never did they have access — they did. It is was this use of it consistent with their work.

Access and activity signals:

  • Records viewed with no corresponding business event: no call, no case, no transaction, no ticket.
  • Searches by name or attribute rather than by the identifier a genuine enquiry would use — browsing rather than looking something up.
  • Access to high-profile, high-balance or related-party accounts with no relationship to the user's portfolio.
  • Activity outside normal hours or patterns for that role, particularly during known absences of supervisors.
  • Volume anomalies against the person's own history and against their peer group.
  • Repeated access to the same records over time with no change recorded.
  • Use of shared or service accounts where individual attribution is lost. An unattributable action is a control failure in itself.

Outcome signals — the strongest, and the least used:

  • Accounts opened by a given member of staff that later show mule behaviour, at a rate above their peers'.
  • Alerts closed by a given investigator that are later reopened or proved true.
  • Refunds, write-offs, exceptions or overrides concentrated in one person's hands.
  • Customers served by one employee appearing disproportionately in fraud cases.

Technical and governance basics worth stating: logs that the subject cannot alter; privileged access that is time-bound and reviewed; joiners-movers-leavers discipline, because accumulated access from previous roles is how long-serving staff end up able to do anything; and access removal that happens at the moment of departure rather than at the end of the month.

The fairness boundary, and it is not optional. Two practices in this area are frequently proposed and should be approached carefully: lifestyle monitoring — noticing a car, a holiday, a watch — and broad monitoring of employees' personal circumstances. Both are intrusive, both have a poor evidential record, and both fall under employment and data protection law. There is also a bias problem: what reads as "living beyond their means" is heavily coloured by assumptions about who should have what. The defensible version is narrow, declared and proportionate: conflict-of-interest declarations, gifts and hospitality registers, outside-interest disclosure, and investigation triggered by what the person did at work, not by what they appear to own.

The notice period deserves naming separately. The window between a resignation and a last day is the highest-risk period in most insider cases, and it is usually the period in which access is least changed: the person still needs their systems to hand over, their manager is focused on cover, and nobody wants an awkward conversation with someone who is leaving on good terms.

It is also when data leaves — client lists, pricing, models, contact details, documents mailed to a personal address or copied to personal storage. Much of that is taken by people who do not think of themselves as doing anything wrong.

So decide the position in advance rather than case by case: what access narrows on the day notice is given, what is logged more closely, what a handover may and may not include, and who confirms on the last day that access actually ended.

Lesson 5 of 5

Handling it, and why resignation is the wrong ending

The most common ending to an insider case is a resignation. The employee leaves, the matter is closed, the institution avoids a tribunal and the embarrassment, and a reference is given that says only dates of employment. It is understandable, and it is a failure with four consequences:

  • No report is made, so a predicate offence — and often a tipping-off or facilitation offence — goes unreported.
  • The person moves to another institution, frequently in the same sector and sometimes in the same role.
  • The scheme is not unwound. Nobody traces what else that access touched, so the accounts, the customers and the external parties involved remain live.
  • Staff learn what happens, which is nothing, and the deterrent value of the whole control environment falls.

What a proper handling looks like:

  • Preserve evidence before acting. Access logs, system records, communications, case files. Once a person knows they are suspected, the record is at risk — and in insider cases the person usually has the ability to alter parts of it.
  • Decide reporting separately from discipline. The employment process and the suspicious activity report (SAR; the FATF standards and many other countries call the same filing a suspicious transaction report, or STR) answer different questions and have different thresholds. An employment outcome is not a substitute for a report, and a report does not require a disciplinary conclusion.
  • Control the access immediately, but think about timing. Removing access is also a signal, and in an enabler case it can tip off an external party. This is the same tension as elsewhere in financial crime, and it needs a decision rule written in advance.
  • Trace the blast radius. Every account opened, every alert cleared, every record viewed, every exception granted. The person is one case; what they enabled is the actual exposure.
  • Work with HR as a partner, not a hand-off. Employment law, evidence standards, union or works-council obligations, and the employee's own rights are real constraints. They are not reasons to avoid reporting.
  • Look at the control that failed, not only the person. If one employee could do this alone, the design assumed something that was not true.

On references. Giving a bland reference to someone dismissed for fraud is lawful in most places and is how the problem moves. Where a regulated regime permits or requires accurate regulatory references, use it; where it does not, the honest answer is that the institution should at least report the conduct so that someone else can see it.

The judgement to take away. Insider fraud is the one financial crime where the institution controls nearly every variable — who is hired, what they can reach, what is logged, what is reviewed, and what happens when they are caught. That makes it the area where an institution's actual tolerance for fraud is most visible, and the quiet resignation is the clearest statement of what that tolerance is.

Knowledge check

Twenty questions

Answer all twenty questions, then check your answers. You need 18 out of 20 to pass and receive a certificate. If you score less, you will see which answers were right and wrong, and then go through the course again before you retake the check. Your answers, progress and times are kept only in this browser.

Standards and frameworks covered

The published frameworks this course teaches toward. Saolix is not affiliated with, endorsed by or accredited by any of these bodies; naming a standard is a citation, not a claim about its author.

Sources

The official documents this course relies on. Laws and guidance change, so check the current version.

  1. A Framework for Managing Fraud Risks in Federal Programs (GAO-15-593SP) · US Government Accountability Office
  2. The IIA's Three Lines Model (July 2020) · The Institute of Internal Auditors
  3. AS 2401: Consideration of Fraud in a Financial Statement Audit · Public Company Accounting Oversight Board
  4. GovS 013: Counter Fraud (Government Functional Standard) · Cabinet Office (UK)
  5. SP 800-53 Rev. 5: Security and Privacy Controls for Information Systems and Organizations · National Institute of Standards and Technology
  6. Financial Trend Analysis: Identity-Related Suspicious Activity (January 2024) · Financial Crimes Enforcement Network (FinCEN), US Treasury
  7. Illicit Financial Flows from Cyber-enabled Fraud (November 2023) · FATF, Egmont Group and INTERPOL
  8. The FATF Recommendations (Recommendations 18, 20 and 21) · Financial Action Task Force
  9. Implementing Phishing-Resistant MFA (fact sheet) · US Cybersecurity and Infrastructure Security Agency