The shape every scam shares
The typologies differ in their cover story and almost nothing else. Underneath, a scam is a sequence, and knowing the sequence is what lets an institution intervene at a useful moment rather than after the last payment.
1. Contact. A message, a call, a dating profile, a social media advertisement, a job offer, a wrong-number text that turns into conversation. Contact is cheap and is made at enormous volume; the scammer is not selecting a victim so much as waiting for one to answer.
2. Trust. Days to months of ordinary conversation, with no money mentioned. This is the part institutions and families underestimate: by the time money is discussed, the relationship is real to the victim and often the most attentive one in their life.
3. The hook. A small, winnable opportunity. A modest investment that returns. A loan repaid. A favour that goes well. The first transaction is designed to succeed, because its job is to convert trust into belief.
4. Escalation. Larger sums, faster. Savings, then pension drawdown, then borrowing, then selling property. The scammer is now coaching the victim, including on what to tell their bank.
5. Isolation. Friends and family are reframed as not understanding; the bank is reframed as an obstacle or a rival. The victim is given scripts for the questions staff will ask. This is why branch and call-centre questioning so often fails: the victim has been prepared for it, by someone they trust more than you.
6. Extraction and the second scam. Withdrawals are blocked by a fee, a tax, a compliance charge, a lawyer's cost — each requiring more money. Then, often, a "recovery" contact arrives offering to get the lost funds back for a fee. Victims are re-victimised at a strikingly high rate, by the same networks selling their details on.
The word that matters: authorised. In each case the victim instructs the payment themselves, knowingly, from their own account. Every control built around unauthorised use — stolen cards, account takeover, device anomaly — sees nothing wrong, because nothing is wrong technically. The customer intends the payment. What is false is what they believe about it.
Which is why the intervention point is not the payment. It is the pattern around it: the new beneficiary, the unusual destination, the account drained in a sequence rather than a transaction, the loan taken to fund a transfer, the customer whose answers are fluent and rehearsed.
One shape breaks that sequence, and it is the one institutions see most often. The purchase scam has no trust phase at all: an advertisement for something that does not exist — concert tickets, a car, a puppy, a holiday let, a phone — and a request to pay by bank transfer because the marketplace's own payment is "not working" or there is a discount for paying directly.
Stages two to five are simply absent. There is one contact, one payment, and no further conversation. Individually the sums are small, which is why these cases attract little training attention and almost no press, and collectively they are the bulk of reported scams.
Two things make them worth separating. The payment method decides who bears the loss: the same fake listing paid by card is a chargeable dispute, and paid by transfer is usually not. And the only reliable intervention is before the money moves — asking a customer paying a stranger for goods whether they have seen the item, and whether they were asked to leave the platform to pay.
Investment fraud, and the compounds behind it
The long-con investment scam is the most lucrative shape and the best documented. It is often described by the term criminals themselves use, which refers to fattening an animal before slaughter — a phrase worth knowing because it tells you the model, though it says something ugly about the victim and some institutions prefer "long-con investment fraud".
How it runs. Contact through a dating app, social platform or a misdirected message. Weeks of conversation. The scammer mentions trading casually, as part of their own life, not as a pitch. The victim asks. A platform is introduced — a convincing application or website, sometimes a clone of a real broker — and a small deposit is made.
Why the platform shows a profit. Because the numbers are typed in. The platform is not connected to any market; it is a display. The victim watches their balance rise, withdraws a small amount successfully — which is paid out of someone else's deposit — and concludes it is real. That single successful withdrawal does more work than every other element combined.
How it ends. The victim invests progressively more, often borrowing. When they try to withdraw a large sum, a fee appears: tax, an anti-money-laundering charge, an account upgrade, a lawyer. Each is paid in hope. The platform eventually stops responding, or the account is shown as liquidated.
The crypto layer. Funds are moved into crypto-assets, by the victim themselves at the scammer's instruction, often through a legitimate exchange and then to a wallet the scammer controls. Two consequences: the victim has performed the conversion, which complicates recovery and sometimes reimbursement; and the proceeds reach a settlement layer where tracing is possible but return is not. The FATF's targeted updates on virtual assets set out how unevenly the obligations that would help are implemented across jurisdictions.
Who is on the other end. This is the part most training omits. A significant share of these operations run from compounds where the workers are themselves trafficked — recruited with fake job advertisements, held under debt bondage, with quotas and punishment. The person typing is frequently a victim of a different crime. It changes nothing about how an institution protects its customer, and it changes a great deal about how the problem should be described, and about why law enforcement treats these as organised crime rather than as isolated frauds.
Adjacent shapes worth recognising: pump-and-dump schemes in thinly traded assets, promoted through social channels and then sold into the demand they create, on which the US Commodity Futures Trading Commission has published plain customer guidance; fake initial offerings; and "recovery" services sold to people who have already lost money.
Romance, and where it meets mule recruitment
Romance fraud is not a category of investment fraud, although the two now overlap heavily. It deserves separate treatment because the harm is different and because the financial signatures differ.
The pattern. A profile that is attractive but not implausible. Rapid intensity — daily contact, affection, future plans — with persistent reasons never to meet or appear on video. An occupation that explains both the distance and the money: offshore work, military deployment, international medicine, an engineering contract.
The asks, in order. They escalate and each has a story: a travel cost, a customs or visa fee, a medical emergency, equipment seized, a business opportunity the victim can share in. Later asks often migrate towards investment, which is where romance and the long con converge.
What makes it hard for an institution. The victim is not confused and is frequently articulate and insistent. They will state plainly that this is their partner and that the money is theirs to send. Staff who press are experienced as intrusive and are often reported as such. And unlike investment fraud, the money frequently goes to ordinary domestic accounts rather than to an exchange or abroad, because the recipient is a mule.
Where it becomes mule recruitment. The same relationship is used to ask the victim to receive and forward money: "my company cannot pay me here", "hold this for us", "send it on for me". The victim becomes a money mule without ever regarding themselves as anything but helpful. Three consequences follow, and institutions handle them badly:
- The victim's account shows classic mule behaviour — credits from unrelated people, immediate onward transfer.
- Standard mule handling closes the account and records a fraud marker, which can cost that person banking access for years.
- The person is simultaneously a victim of romance fraud and the operator of a mule account, and both are true.
So the handling rule. Where mule behaviour appears alongside a relationship the customer describes, treat the possibility of coercion or deception as live — exactly as in the trafficking and mule material elsewhere in this catalogue — and route it through whatever vulnerable-customer process exists, rather than straight to closure. Report the activity; separate that from a judgement about the person.
Impersonation, and what AI changed
The core deception. The caller or message appears to be an authority the victim already trusts: their bank's fraud team, the police, a tax authority, a utility, a delivery company, a government benefits office.
The most damaging version — the "safe account". The victim is told their account is compromised and that they must move their money immediately to a secure account "while we investigate". It is effective because it uses the bank's own language, arrives with urgency, and gives the victim a reason to lie to the branch: they have been told the staff member may be the problem.
Other recurring forms:
- Tax and government impersonation, usually with a penalty and a deadline.
- Invoice redirection and business email compromise, where a supplier's bank details appear to change. The damage per case is the largest of any scam type, and it targets finance teams rather than consumers.
- The family emergency, in which a relative is apparently in trouble and needs money now. This is the form artificial intelligence has changed most: voice can be cloned from a short public sample, so the phrase "it sounded exactly like my daughter" is no longer evidence of anything. The US Federal Trade Commission has published consumer guidance on precisely this, and the FBI has warned that criminals use generative AI to make these approaches more convincing at scale.
- Job and task scams, where small payments are received for simple online tasks before the victim is asked to deposit their own money to unlock higher earnings.
What AI changed, stated carefully. Not the structure — every stage in lesson one is unchanged. What it changed is cost and plausibility: fluent messages in any language with no errors to notice, cloned voices, convincing video, and automation that lets one operator run many conversations. The practical consequence for training and for public advice is that the old tells — poor spelling, odd phrasing, refusal to speak — are no longer reliable, and advice built on them is now actively misleading.
What still works as advice. Verify through a channel the other party did not give you: a number you already had, an application you opened yourself, a person you call back. That advice survives voice cloning, because it does not depend on recognising anything.
A note on the business victim, because the controls are different. Invoice redirection does not target a person's savings; it targets a process. The attacker does not need to deceive a victim about reality, only to get one field changed in a supplier record.
So the control is not fraud awareness, it is change control over supplier bank details:
- Verify a change of bank details by calling a number you already hold for that supplier, never one in the email requesting the change, and never by replying to it.
- Authorise the change, not just the payment. Most finance teams have dual authorisation on payments and single authorisation on static data, which is the wrong way round: the payment looks ordinary precisely because the record was already altered.
- Make the first payment after any details change a checkpoint, and tell the supplier you will do that, so it is expected rather than awkward.
- Expect the request to arrive at the worst moment — a deadline, a quarter end, an executive apparently travelling. Urgency from a senior name is the mechanism, not a reason to skip a step.
What an institution can see, and what actually helps
Before the payment. None of these is conclusive and all are checkable:
- A first payment to a new beneficiary that is large relative to the account's history.
- A sequence rather than a transaction: several payments over days, rising in size.
- Savings or a term deposit broken, a pension drawn down, or a loan taken shortly before a transfer.
- Payments to a crypto exchange from a customer with no history of them, especially an older customer.
- A customer in branch or on the telephone who is fluent, insistent and slightly rehearsed, who declines to explain, or who is receiving messages during the conversation.
- A change of contact details shortly before the activity begins.
What the evidence says about warnings. Generic warning screens are weak: customers click past them, and the scammer has usually pre-empted them. What performs better is specific, human and timed — naming the exact scam shape that matches the pattern, delivered by a person, with a pause rather than a block. Effectiveness is testable, and an institution that has never measured which of its interventions stop a payment is guessing.
On blocking and friction. Holds work and are resented. The cost is real: a wrongly delayed payment to a genuine builder, a missed house purchase. The honest position is that some friction is correct for a first large payment to a new beneficiary, and that the institution should be able to say what proportion of its holds turned out to be scams.
Treating the person properly. Scam victims are routinely described, including internally, as having been careless. That is wrong on the evidence — these are engineered deceptions run by organised networks, and the victims are not unusually gullible — and it is operationally harmful: a customer who expects to be judged will not tell you the truth about why they are sending money, which removes your last chance to intervene.
After the event. Contact the receiving institution immediately, because recovery depends on minutes; report the suspicion, naming the typology, since the receiving account is almost certainly operating for more than one victim; record the customer as at risk of re-victimisation, and expect a recovery-scam approach to follow; and feed the case into monitoring, because the beneficiary, the device and the destination are all reusable intelligence.
The measure that matters. Not warnings shown. How many payments were stopped, how much was recovered, and what proportion of customers who were warned still sent the money — because that last number is the only honest test of whether the intervention works.
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.
Your answers
Your certificate of completion
Enter your name as you want it to appear, then save the certificate as a PDF. In the print window, choose Save as PDF. A certificate is issued once per completion of the course.
Rate this course
How would you rate this course? Your rating is anonymous — just a number, with no name and no account.
Bring this to your team
Your colleagues can take this course too — it is free, needs no sign-up, and nothing they enter leaves their browser. If your institution wants the same ground covered as structured training for a team, with an instructor and your own cases, we run courses in a classroom or online for groups of ten or more.
Talk to us about training your team → See the paid courses →
Saolix does not record who takes this course, so it cannot verify these certificates. The certificate confirms completion of a free self-paced course and is not an accredited qualification.
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.
- Internet Crime Report 2025 · Federal Bureau of Investigation, Internet Crime Complaint Center (IC3)
- Illicit Financial Flows from Cyber-enabled Fraud (November 2023) · FATF, Egmont Group and INTERPOL
- Criminals Use Generative Artificial Intelligence to Facilitate Financial Fraud (PSA, December 2024) · Federal Bureau of Investigation, IC3
- Scammers use AI to enhance their family emergency schemes · US Federal Trade Commission
- Customer Advisory: Beware Virtual Currency Pump-and-Dump Schemes · US Commodity Futures Trading Commission
- Seventh Targeted Update on Implementation of the FATF Standards on Virtual Assets and VASPs · Financial Action Task Force
- Financial Flows from Human Trafficking (July 2018) · FATF and Asia/Pacific Group on Money Laundering
- 2025 Report on Payment Fraud (December 2025) · European Banking Authority and European Central Bank
- The FATF Recommendations (Recommendations 1, 16 and 20) · Financial Action Task Force



