Fraud
Account takeover (ATO)
Someone other than the customer controls the account: a login from a new device or place, or a change of phone, email, or password, followed soon after by a payout, a new beneficiary, or a drained balance.
What it catches
A criminal gets into a real customer’s account and uses it as their own. They get in with stolen or phished credentials, a one-time code the customer was tricked into sharing, a SIM swap that moves the customer’s phone number to the criminal’s SIM, or a malware-infected device. Then they act fast: they change the phone number, email or password so the customer cannot get back in or be warned, add their own beneficiary, raise limits, and send the balance out.
FinCEN defines account takeover as a scheme where the customer, not the institution, is the target, with “the ultimate goal … to remove, steal, procure or otherwise affect funds of the targeted customer”, and names as signs “sudden wire transfers, or changes to customer and account profiles” (FIN-2011-A016). The FATF, INTERPOL and Egmont Group list among cyber-enabled fraud indicators “frequent changes of contact details, phone numbers, email addresses after opening of the account”, “multiple failed login attempts”, and “multiple IP addresses or electronic devices associated with a single online account”.
The rule watches the sequence: a session or profile signal that the person in control has changed, then a sensitive action soon after. Either half alone is common and mostly innocent; the pair is the signal.
What it does not catch: a scam where the real customer is persuaded to pay the criminal from their own phone (authorised push payment fraud: see the new-beneficiary rule), a takeover on the customer’s usual device (remote-access malware, a family member), or a slow takeover where the criminal waits longer than the window before moving money. Most of the rule runs on login, device and profile-change events. Where these are not sent to the monitoring system, only the transaction end of the sequence is visible.
How the rule works
Keep, for each customer, the devices, IP addresses and countries they have used, and a log of profile changes. When a sensitive action happens, look back over a short window for a takeover signal. Alert when both are present.
Takeover signals (the first half):
- New device. A login or first use from a device never used on this customer before.
- Contact or credential change. Phone number, email, password or PIN changed or reset, a new device enrolled, or a recent SIM change on the customer’s phone number where your telecom data shows it.
- Location jump. Two sessions from countries too far apart for the time between them (“impossible travel”; neighbouring countries and sessions through a VPN or proxy do not count), or a session from a country that does not fit the customer’s declared country.
- Device used by other customers. The new device is already linked to other, unrelated customers (see shared identifiers).
Sensitive actions (the second half): a new beneficiary added, a transaction limit raised, an outgoing payment above your floor, or a payment that takes most of the balance.
Looks at
Every sensitive action on a customer’s account: a new beneficiary added, a transaction limit raised, or an outgoing payment.
Alerts when
Both of these are true:
- At least one takeover signal is present (see below).
- The action is a new beneficiary, a raised limit, an outgoing payment of at least the floor of your new-beneficiary rule, or an outgoing payment that takes at least 98% of the balance before it and is at least your small minimum.
The floor and the small minimum are set per segment and market, and payments are converted to one reference currency to compare with them. Where you can, hold the payment while the alert is reviewed.
Counts as a takeover signal
Any of these. The first three count only if they happened in the 24 hours up to the action:
- New device: the customer logged in or acted from a device they had never used before.
- Contact or credential change: the customer’s phone number, email, password, PIN or enrolled device was changed, or, where your telecom data shows it, the SIM behind their phone number was changed.
- Location jump: two of the customer’s sessions came from countries too far apart for the time between them (neighbouring countries and sessions through a VPN or proxy do not count), or a session came from a country that does not fit the customer’s declared country.
- Device used by other customers: the device behind the action is also linked to other customers who are not related to this one.
The alert shows
The action, and every takeover signal that was found.
Highlighted values are settings you choose. Hover over one, tap it or tab to it for a suggested value.
Combination window How long before the action a takeover signal still counts. Start at 24 hours. A longer window catches criminals who wait but pairs more unrelated events. Set it from the timing of your own confirmed takeover cases.
Amount floor The smallest outgoing payment that counts as a sensitive action after a takeover signal. Start at the floor of your first-payment-to-new-beneficiary rule, set per market and segment. Lower it where losses are small and frequent, as in consumer wallets; raise it for business accounts that change devices often.
Drain share How much of the balance before the payment the payment must take. Start at 98%, which catches near-full drains. A lower share also catches criminals who leave a little behind, but adds customers who simply spend most of their balance.
Small minimum The smallest payment the drain test counts, so that near-empty balances do not alert. Set it below the amount floor, from your own distribution of payment amounts.
Run it when the action is requested, before the money leaves: a takeover alert after settlement is mostly a recovery case.
Data it needs
ShowHide
| Field | Usually comes from | If it’s missing |
|---|---|---|
| Login or session event: customer, timestamp | App or online banking backend | The rule cannot see who is in control. Only the transaction half runs; use the new-beneficiary and dormant-account rules instead. |
| Device identifier on each login | Mobile app or device SDK | No “new device” signal. Check which identifier stays the same across sessions and app updates; an identifier that changes on every install makes every login look new. |
| IP address and IP country on each login | Backend, geolocation service | No location jump. Keep the session country in its own field; never let it overwrite the customer’s declared country. |
| Profile-change event naming the field changed (phone, email, password, PIN, device) | Customer profile service | A “profile updated” event that does not say what changed cannot separate a new phone number from a new nickname. |
| Outgoing payments: amount, currency, counterparty, timestamp, status | Ledger, payment processor | Required for the second half. |
| Balance before each payment | Ledger | No drain test; the amount floor still runs. |
| Beneficiary-added and limit-change events | Beneficiary and limit management | Those two sensitive actions are invisible. |
| SIM-change date for the customer’s phone number (optional) | Mobile network operator data, where available | You lose the SIM-swap signal; the contact-change and new-device signals still run. |
Device cookies, geolocation and IP matching on their own are weak evidence: the FFIEC calls them “insecure and ineffective if used alone” for authentication, and useful only combined with other controls (2021 guidance, Appendix). The same holds for monitoring, which is why the rule pairs them with a sensitive action.
Thresholds
Each row is a setting the rule needs. Set by law values come from the source named under “Where it comes from”. Suggested values are starting points: test them on your own history, then adjust.
| Parameter | Value | Type | Where it comes from |
|---|---|---|---|
| Combination window (takeover signal, then sensitive action) | 24 hours | Suggested | A longer window catches criminals who wait but pairs more unrelated events (a customer who changed phones last week and pays rent today). Set it from the timing of your own confirmed takeover cases. |
| Amount floor for an outgoing payment after a signal | The floor of your first-payment-to-new-beneficiary rule, set per market and segment | Suggested | A takeover signal lowers the bar for review; it does not need its own high floor. Lower it where losses are small and frequent (consumer wallets), raise it for business accounts that change devices often. |
| Drain share of balance | 98% of the balance before the payment | Suggested | Catches near-full drains. A lower share also catches criminals who leave a little behind, but adds customers who simply spend most of their balance. Pair it with a small minimum amount, below the amount floor and set from your own distribution, so near-empty balances do not alert. |
| Location jump | Sessions from two countries that do not border each other within 1 to 3 hours | Suggested | A longer gap catches more but adds travellers. Leave out sessions flagged as VPN or proxy: their country is the exit server’s, not the customer’s. |
| Device history before “new device” means anything | At least your customers’ usual gap between logins | Suggested | At go-live every device is new. Run in observation mode until the history covers the gap, and longer where customers log in rarely. |
| Nigeria: outflow limit on a newly activated mobile app (existing accounts) | An outflow transaction limit set by the institution, at most ₦20,000, in the first 24 hours after activation | Set by law | CBN circular PSP/DIR/PUB/CIR/001/001, 12 March 2026, item 5(iv), in force from 1 July 2026. A control the institution must impose, not an alert threshold. The circular calls ₦20,000 a “maximum transaction limit” and does not say whether it applies per transaction or in total over the 24 hours. Moving the app to another device triggers re-activation (item 5(ii)); our reading is that the limit then applies again. A rule can check the control holds: an outflow above the limit within 24 hours of an app activation means it failed. |
Apart from the Nigerian limit, no regulator we reviewed sets a number for this rule. FinCEN, the FATF, the FFIEC and the EU technical standards name the signals, not thresholds.
By country
Nigeria. The CBN’s March 2026 instant payment circular requires that mobile financial services apps “shall only be enabled on one device at a time”; migration to another device “shall trigger automatic re-activation and authentication”; outflows on an existing account’s newly activated app are capped for the first 24 hours (see Thresholds); and “first-time log on a new device shall require additional MFA” for internet banking. It also requires enterprise fraud monitoring “for both inflows and outflows” and enhanced authentication for online account reactivation (items 3, 4 and 5). The CBN’s 2024 cybersecurity framework names SIM swap among the cyber-risks of USSD banking (section 4.1.3). A suspicious transaction is reported within 24 hours (see the reporting deadline below). For stablecoin issuers and virtual asset service providers (VASPs) licensed by the CBN, its Regulations for the Operations of Non-Security Virtual Assets in Nigeria (in force 5 November 2026) require authentication “proportionate to transaction and account risks” that protects “against credential theft and unauthorised transfers” (section 37.11), require the provider to “detect repeated failed logins and anomalous sessions, apply controls against automated attacks and account takeover and protect session credentials”, with “risk-based session termination and re-authentication controls” (section 37.12), and require monitoring of “suspicious device, access, wallet and transaction patterns” (section 37.13).
Reporting deadline
Report to the NFIU immediately, with a written report within 24 hours (Money Laundering (Prevention and Prohibition) Act 2022, s.7(1)–(2)). The NFIU’s December 2024 STR guidelines start the 24 hours once the transaction is found suspicious, after an examination of no more than 72 hours, and require written reasons for every alert closed as not suspicious (STR-NFIU-2024-A0001, Part Six, para 2).
Ghana. The Bank of Ghana’s Cyber and Information Security Directive (March 2026) requires “an automated system for detecting and monitoring anomalies in customer accounts, particularly when high-risk transactions may be underway”, immediate customer alerts on unusual activity, and approval requests “through a different communication channel or device other than those being used for the transaction in question” (para 74(1)–(4)). GIABA’s 2025 cybercrime report describes two Ghanaian cases. In one, criminals holding customer details from insiders at some banks phoned customers who were being enrolled in electronic banking, from numbers made to look like the bank’s, tricked them into giving their log-in details, and made unauthorised withdrawals through mobile and internet banking, with funds “immediately withdrawn from Mobile Money Merchants”. The other was a SIM swap plot in which a telecom employee cloned victims’ numbers so the criminals could try to withdraw from their bank accounts, stopped when a bank official reported it (Typology 6, Cases 4 and 5). The BoG/FIC AML/CFT/CPF Guideline 2025 makes fraud part of the AML/CFT/CPF regime: institutions licensed by the Bank of Ghana must “ensure that effective systems and policies are in place to detect and curb fraud activities” (2.14(2)), “acquire or develop a fraud detection tool” embedded at a minimum with threshold limits, the customer profile and the features of products and services (2.14(4)(ii) and (iii)), and report any fraud incident to the BoG immediately on detection and file a fraud report through ORASS as and when fraud occurs (2.14(4)(vii)). They must “Implement multifactor authentication and liveness check for all transactions on mobile banking application during onboarding” (2.4.2(3)(ii)) and verify customers during onboarding and ongoing transactions with biometric, liveness and multifactor authentication tools where applicable (2.6.3(4)(vi)).
Reporting deadline
Within 24 hours of forming the suspicion, attempted transactions included, whatever the amount (FIC Ghana, “Report Types”; BoG/FIC AML/CFT/CPF Guideline 2025, 2.8.1(6) and 2.8.3(2)).
West Africa (GIABA). GIABA lists among mobile money fraud methods scammers who “impersonate customer service agents and ask users to share PINs, OTPs (One-Time Passwords), or account details” and “complex social engineering schemes to obtain log-in credentials” (Typology 6, p.40).
WAEMU (UEMOA). The BCEAO’s payment services instruction requires strong customer authentication when the payer accesses the account online, initiates a payment, or “exécute toute autre action susceptible de comporter un risque de fraude en matière de paiement” (carries out any other action that may carry a payment fraud risk) (Instruction n°001-01-2024, art. 85). For an unauthorised payment, the provider must “rembourser immédiatement” (refund immediately) (art. 69), unless the payer acted fraudulently or failed one of their own obligations, which include taking “toute mesure raisonnable” (every reasonable step) to keep the payment instrument secure (arts. 65 and 70). The provider must prove that the payment was authenticated, and its record that the instrument was used is not enough on its own to prove the payer authorised it (art. 68). WAEMU (UEMOA in French) is the West African Economic and Monetary Union: Benin, Burkina Faso, Côte d’Ivoire, Guinea-Bissau, Mali, Niger, Senegal and Togo, the same eight states that form the West African Monetary Union (UMOA). The BCEAO’s instructions and the Council of Ministers’ decisions cited here apply in all eight. The uniform law applies in each state through that state’s own act, which may number its articles differently. The law of a single state is covered only where that state is named.
Reporting deadline
Report to the CENTIF “immédiatement” (Loi uniforme 2023, art. 60). The law defines “sans délai” as at most 24 hours (art. 2, point 58); “immédiatement” is not defined, and our reading is to treat 24 hours as the outer limit. Do not carry out an operation you suspect until the report is made; if it has already gone through, inform the CENTIF “sans délai” (art. 61).
EU. The SCA technical standards require payment service providers to run transaction monitoring that takes into account, at a minimum, “lists of compromised or stolen authentication elements”, “known fraud scenarios”, “signs of malware infection in any sessions of the authentication procedure” and, where the provider supplied the access device or software, “the abnormal use of the access device or the software” (Delegated Regulation (EU) 2018/389, art. 2(2)). A remote payment may skip strong authentication as low-risk only if real-time risk analysis found no “unusual information about the payer’s device/software access” and no “abnormal location of the payer”, among other conditions (art. 18(2)(c)). PSD2 requires strong authentication when the payer “carries out any action through a remote channel which may imply a risk of payment fraud or other abuses” (Directive (EU) 2015/2366, art. 97(1)(c)), and the standards require it when a payer “creates or amends a list of trusted beneficiaries” (art. 13(1)). In the EBA and ECB’s 2025 payment fraud report, most fraud by value on card payments, cash withdrawals and e-money was payments the fraudster issued (more than 90% in each of the six periods analysed, except e-money in 2024), while most credit transfer fraud came from manipulating the payer (74% by value in 2024) (section 3.2). The report does not say how the fraudster came to issue the payment: account takeover is one way, and card payments with stolen card details, which a login, device and profile rule does not see, are another. Fraud from manipulating the payer needs the new-beneficiary and scam checks.
Reporting deadline
Inform the national FIU “promptly” (Directive (EU) 2015/849, art. 33(1)). The directive sets no number of hours; your member state’s law may.
United States. FinCEN asks filers to use the term “account takeover fraud” in the SAR narrative (FIN-2011-A016) and to include “IP addresses with timestamps” and “device identifiers” in SARs (FIN-2016-A005). The current SAR form has an “Account takeover” box (item 38(a)) and a field for the IP address used to log in to online or mobile banking, with its date and time, to be explained in the narrative (item 43) (FinCEN SAR Filing Instructions, April 2026). Its 2024 deepfake alert lists “access to an account from an IP address that is inconsistent with the customer’s profile” and “a customer’s geographic or device data is inconsistent with the customer’s identity documents” as red flags (FIN-2024-Alert004). FinCEN’s 2024 identity analysis counted about 80,000 account takeover reports in 2021 BSA filings, about $9 billion in suspicious activity. The FFIEC’s 2021 authentication guidance, which gives examples and imposes no new requirements, lists among examples of controls enhanced authentication for “changes to physical or email address, password, contact information, or enrolled devices” and fraud monitoring for “changes in user or customer behavior or transaction velocity and increases in login or account lockout activity”; for call centres it lists controls that “identify deviations from a customer’s or user’s usual geographic location” of callers and processes for “changes to established phone numbers or carriers” (Appendix).
Reporting deadline
No later than 30 calendar days after initial detection (31 CFR 1020.320 for banks, 1022.320 for money services businesses). A bank that has identified no suspect may take 30 more days to identify one, never more than 60 days in all (1020.320(b)(3)); the rule for money services businesses has no such extension.
Argentina. Among the circumstances to weigh in alerts: situations where a person “suplantare, se apoderare o intentare suplantar la identidad de una persona humana sin su consentimiento” (impersonated, took over or tried to impersonate the identity of a natural person without their consent), and an account “asociada a varios dispositivos electrónicos, sin justificación aparente” (linked to several electronic devices without apparent justification) (UIF Resolution 200/2024, art. 34(b), circumstances 9 and 15).
Reporting deadline
Money laundering, within 24 hours of concluding that the operation is suspicious, and no later than 90 calendar days from the operation, performed or attempted; terrorist financing and proliferation financing, within 24 hours of the operation (UIF Resolution 200/2024, art. 36(c)).
False positives and tuning
ShowHide
Common innocent explanations: a new phone, a reinstalled app or a factory reset; travel, roaming and VPNs; a family sharing one phone; a customer who changes phone number and then pays rent the same day; a business whose staff log in from several devices.
Typical causes of misfires:
- Unstable device identifiers. Some identifiers change on every app install or update, and an app can send two different device fields that do not agree. Every login then looks like a new device. Before go-live, measure how often a known customer appears with a new identifier and pick the one that stays put.
- Placeholder identifiers. The reverse trap: one value shared by many unrelated customers. On Apple devices, the advertising identifier is all zeros (00000000-0000-0000-0000-000000000000) when the user has not authorised tracking, and in the simulator. Used as a device identifier, it makes many unrelated customers share one “device”, so “device used by other customers” fires on all of them and “new device” never does. Leave out empty and constant values, and do not use an advertising identifier as the device identifier.
- Short device history. At go-live, every device is new and every customer looks taken over. Run in observation mode until the history covers your customers’ usual gap between logins. Customers coming back after a long break on a new phone will still look new: pair the signal with the sensitive action rather than alerting on the device alone.
- “First seen” is not “changed”. A test on when the monitoring system first saw an email or phone number fires on every new customer and on established customers whose data was loaded late. Detect a change by comparing with the customer’s previous value, from a profile-change event.
- Profile events that do not say what changed. If the “profile updated” event carries no field name, a nickname edit and a new phone number look the same. Ask for the field changed, and the old and new values (or hashes of them).
- Location from IP addresses. Mobile carrier and corporate IPs often geolocate to a capital or a neighbouring country, and VPNs show the exit server’s country. Leave VPN and proxy sessions out of the location test, and do not count a move to a neighbouring country as a jump.
- Session location overwriting the profile. If the session’s country is written onto the customer’s country field, a “location does not fit the declared country” check compares a value with itself and never fires. Keep the two in separate fields.
- Duplicate events. Integrations that resend the same login or profile event several times inflate login counts and can make one change look like many. De-duplicate on the event’s own identifiers and timestamp.
- Signals your data cannot supply. A failed-login count needs login outcomes in the data; where they are not sent, the term finds nothing and the rule stays quiet. Test each signal on a known case before relying on it, and check that a quiet rule is actually evaluating events.
- Alerting on the signal alone. A new device or a changed email on its own is mostly customers changing phones. Alert on the combination; use the single signal to raise the priority of other alerts.
How to tune: start from the combination, not the single signals; set the window from your confirmed cases; exclude VPN sessions from the location test; and measure the new-device rate on known-good customers before going live.
When it fires
ShowHide
Analyst checklist:
- Hold the payment or the change if it has not gone through.
- Contact the customer through a contact detail on file before the change, not one supplied in the session. A criminal who changed the phone number will answer on the new one.
- List the logins, devices, IP addresses and countries for the past weeks. Note when the new device first appeared and what was done from it: contact changes, new beneficiaries, limit changes, payments.
- Check whether the new device or IP is linked to other customers (see shared identifiers) and whether the beneficiary is paid by other customers or was reported by another institution.
- If the customer confirms they did not act: secure the account (end sessions, reset credentials, revert contact changes from a channel on file), recall payments, and tell the receiving institution.
- If the customer did act but under someone’s instructions, treat it as a scam or a mule recruitment, not a takeover.
The FATF notes that a single indicator “may not alone warrant suspicion”. An alert is a reason to review, not a suspicion in itself: the NFIU’s 2024 STR guidelines call monitoring alerts “unusual” and say they “should not be automatically regarded as suspicious” (Part Three, paras 1–2). File a suspicious transaction report when the review gives grounds to suspect. A payment the rule stopped can still need one: the reporting duty covers attempted transactions in Nigeria (Money Laundering (Prevention and Prohibition) Act 2022, s.7(3): “whether the transaction is complete or not”), the UEMOA zone (“tentatives d’opérations”, Loi uniforme 2023, art. 60), the EU (Directive (EU) 2015/849, art. 33(1)) and the United States (“conducted or attempted”, 31 CFR 1020.320(a)(2) and 1022.320(a)(2)). In the United States, a SAR is required from USD 5,000 for banks and USD 2,000 for money services businesses (same paragraphs); check “Account takeover” (item 38(a)), use the term “account takeover fraud” in the narrative and include IP addresses and device identifiers (see By country). Each country’s reporting deadline is there too.
Related rules
- New beneficiary, high-value payment: the payment half of the same sequence, and the rule for scams where the customer pays.
- Shared identifiers across customers: a new device already used by other customers.
- Dormant account reactivation: a takeover of a quiet account, often after a phone or contact change.
- Transaction velocity: many quick payments out after a login from a new device.
- Rapid pass-through: the drain, when stolen money passes through a mule account.
- New account activity: the same behaviour on an account opened with stolen identity details rather than taken over.
- Sudden change from a customer’s baseline: spending far out of the customer’s pattern.
- Card cycling and cross-merchant card use: many different cards added to one account in a short time, sometimes right after a new device or a changed email or phone.
- Refused or failed identity verification, and false KYC data: a change to the customer’s name or date of birth, or identity data that fail a check, rather than a change of contact details.
- High-risk jurisdictions and country mismatch: a card issued in one country paying in another, or sessions from a country that does not fit the customer’s declared residence, which can mean someone else is using the account.
- Insider risk (staff in reported cases or acting outside their role): a customer’s details changed by a member of staff in the back office, not by someone who logged in.
- Third-party funding and payment account name mismatch: a payout to an account in a name other than the customer’s, whether or not anything changed on the account first.
Sources
25 sourcesHide
- FinCEN, Account Takeover Activity, Advisory FIN-2011-A016, 19 December 2011, sections “Identifying Account Takeover Activity” (“unusual ATM activity, clustered Automated Clearing House (ACH) transactions in different geographic areas, sudden wire transfers, or changes to customer and account profiles”; “the ultimate goal is to remove, steal, procure or otherwise affect funds of the targeted customer”) and “Suspicious Activity Reporting” (“should use the term ‘account takeover fraud’ in the narrative section of the SAR”). Supports the pattern and the reporting practice. https://www.fincen.gov/sites/default/files/advisory/FIN-2011-A016.pdf
- FinCEN, Advisory to Financial Institutions on Cyber-Events and Cyber-Enabled Crime, FIN-2016-A005, 25 October 2016, p.6 (“Cyber-related information includes, but is not limited to, IP addresses with timestamps, virtual-wallet information, device identifiers, and cyber-event information”). Supports the reporting practice. https://www.fincen.gov/sites/default/files/advisory/2016-10-25/Cyber%20Threats%20Advisory%20-%20FINAL%20508_2.pdf
- FinCEN, Identity-Related Suspicious Activity: 2021 Threats and Trends, Financial Trend Analysis, January 2024, p.6 (“bad actors engage in an ‘Account Takeover’ using stolen authenticators and credentials to gain full access to victim financial accounts”) and the typology table, p.13 (Account Takeover: “~80,000” reports, “$9 billion”). Supports the pattern. https://www.fincen.gov/sites/default/files/shared/FTA_Identity_Final508.pdf
- FinCEN, FinCEN Alert on Fraud Schemes Involving Deepfake Media Targeting Financial Institutions, FIN-2024-Alert004, 13 November 2024, indicators (“Access to an account from an IP address that is inconsistent with the customer’s profile”) and red flags (“A customer’s geographic or device data is inconsistent with the customer’s identity documents”). Supports the pattern. https://www.fincen.gov/sites/default/files/shared/FinCEN-Alert-DeepFakes-Alert508FINAL.pdf
- FFIEC, Authentication and Access to Financial Institution Services and Systems, August 2021, section 2 (“commit fraud and account takeover”) and Appendix: “Device Identification and Enrollment”, “Account Maintenance Controls”, “Identify Unauthorized Access Attempts”, “Lost, Stolen, or Changed Information and Devices”, “Fraud and Anomaly Detection Monitoring”. Supports the pattern and the controls. Copy issued with OCC Bulletin 2021-36: https://www.occ.gov/news-issuances/bulletins/2021/bulletin-2021-36a.pdf
- FATF, INTERPOL and Egmont Group, Illicit Financial Flows from Cyber-Enabled Fraud, November 2023: para 36 (deepfakes and “account takeover fraud”); Annex A preamble (“The existence of a single indicator … may not alone warrant suspicion”); Annex A, pp.64–65 (“Frequent changes of contact details, phone numbers, email addresses after opening of the account”; “multiple failed login attempts”; “Multiple IP addresses or electronic devices associated with a single online account”; “Transactions conducted with device time zone mismatch”; VPNs and remote desktop access); Annex B, pp.67–68 (a “cooling-off period for first time enrolment of online banking services or secure devices”; multi-factor authentication for “adding or activating beneficiaries”; “Monitoring IP addresses used to connect to on-line banking websites”). Supports the pattern. https://www.fatf-gafi.org/content/dam/fatf-gafi/reports/Illicit-financial-flows-cyber-enabled-fraud.pdf
- Commission Delegated Regulation (EU) 2018/389 (regulatory technical standards for strong customer authentication), 27 November 2017, art. 2(1)–(2), art. 13(1) and art. 18(2)(c), as adopted. Supports the pattern. https://www.legislation.gov.uk/eur/2018/389/contents/adopted
- Directive (EU) 2015/2366 (PSD2), art. 97(1), as adopted. Supports the authentication duty for risky remote actions. https://www.legislation.gov.uk/eudr/2015/2366/article/97/adopted
- EBA and ECB, 2025 Report on Payment Fraud, EBA/REP/2025/40, December 2025, section 3.2, p.19 (“In the vast majority of fraudulent card payment transactions, cash withdrawals and e-money transactions, the payment order was issued by the fraudster”; credit transfers “from 65% to 74% in value terms” from manipulation of the payer). Supports the EU context. https://www.ecb.europa.eu/press/intro/publications/pdf/ecb.ebaecb202512.en.pdf
- Central Bank of Nigeria, Additional Functionalities for Instant Payment, circular PSP/DIR/PUB/CIR/001/001, 12 March 2026, items 3, 4 and 5(i)–(v). Supports the Nigerian controls and the legal limit. https://www.cbn.gov.ng/Out/2026/CCD/IP%20Circular.pdf
- Central Bank of Nigeria, Risk-Based Cybersecurity Framework and Guidelines for Deposit Money Banks and Payment Service Banks, 31 May 2024, section 4.1.3 (“Cyber-risks associated with the use of USSD include smishing, social engineering, Distributed Denial of Service (DDoS) and SIM Swap”). Supports the pattern. https://www.cbn.gov.ng/Out/2024/BSD/CBN%20Risk-Based%20Cybersecurity%20Framework%20for%20DMBs%20and%20PSBs_2024.pdf
- Bank of Ghana, Cyber and Information Security Directive, March 2026, para 74 “Customer Security” (1)–(4). Supports the monitoring and out-of-band confirmation duties. https://www.bog.gov.gh/wp-content/uploads/2020/06/Cyber-Information-Security-Directive-2026.pdf
- GIABA, Typologies of Money Laundering and Terrorist Financing Linked to Cybercrime in West Africa, Regional Report, May 2025, Typology 6, Cases 4 and 5 (Ghana), p.38, and “Methods and techniques”, p.40. Supports the pattern. https://www.giaba.org/typologies/3495__typologies-of-money-laundering-and-terrorist-financing-linked-to-cybercrime-in-west-africa.html
- BCEAO, Instruction n°001-01-2024 relative aux services de paiement dans l’Union Monétaire Ouest Africaine, arts. 65, 68, 69, 70 and 85. Supports the authentication duty and the refund rule. https://www.bceao.int/sites/default/files/inline-files/Instruction-No001-01-2024_relative_aux_services_de_paiement_dans_l-UMOA.pdf
- Money Laundering (Prevention and Prohibition) Act 2022 (Nigeria), s.7(1)–(3). Supports the reporting duty, attempts included. https://www.nfiu.gov.ng/LawsAndRegulations
- NFIU, Guidelines for the Identification, Verification and Reporting of Suspicious Transactions Related to ML/FT/PF for Financial Institutions, STR-NFIU-2024-A0001, 13 December 2024, Part Three paras 1–2 and Part Six para 2. Supports the reporting duty. https://www.nfiu.gov.ng/AdvisoryAndGuidance
- UEMOA, Loi uniforme relative à la lutte contre le blanchiment de capitaux, le financement du terrorisme et de la prolifération des armes de destruction massive (March 2023), arts. 2 (point 58), 60 and 61. Supports the reporting duty, attempts included. https://www.bceao.int/sites/default/files/2025-02/Loi-uniforme-LBCFTFP-Mars-2023-vf.pdf
- Financial Intelligence Centre (Ghana), “Report Types” (“filed by Accountable Institutions within twenty-four hours upon the grounds of suspicion of particular transaction”). Supports the reporting duty. https://fic.gov.gh/report-types.html
- 31 CFR 1020.320(a)(2) and (b)(3), and 1022.320(a)(2) and (b)(3) (“conducted or attempted”; “at least $5,000” for banks, “at least $2,000” for money services businesses; “no later than 30 calendar days after the date of the initial detection”; for banks, “an additional 30 calendar days to identify a suspect”, “In no case … more than 60 calendar days”). Supports the reporting duty. https://www.law.cornell.edu/cfr/text/31/1020.320 and https://www.law.cornell.edu/cfr/text/31/1022.320
- FinCEN, FinCEN Suspicious Activity Reporting (SAR) Filing Instructions, Release April 2026, Version 1.0, item 38(a) “Account takeover” and item 43 “IP Address” (“the IP address used to log into the institution’s online banking page or the IP address used to access an institution’s mobile application”; “IP address information must be explained in the FinCEN SAR Part V narrative”). Supports the reporting practice. https://bsaefiling.fincen.gov/resources/FinCENSARFilingInstructions.pdf
- Directive (EU) 2015/849 (fourth anti-money laundering directive), art. 33(1), as adopted (“promptly … informing the FIU”; “All suspicious transactions, including attempted transactions, shall be reported”). Supports the reporting duty. https://www.legislation.gov.uk/eudr/2015/849/article/33/adopted
- Apple Developer Documentation, advertisingIdentifier (AdSupport, ASIdentifierManager): “The advertising identifier returns all zeros (00000000-0000-0000-0000-000000000000)” in the simulator and when the app has not requested, or the user has declined, tracking authorization. Supports the placeholder-identifier trap. https://developer.apple.com/documentation/adsupport/asidentifiermanager/advertisingidentifier
- Unidad de Información Financiera (Argentina), Resolución 200/2024 (RESOL-2024-200-APN-UIF#MJ), 18 December 2024, Boletín Oficial no. 35571, 19 December 2024, art. 34(b), circumstances 9 and 15; art. 36(c) (report deadlines). https://servicios.infoleg.gob.ar/infolegInternet/anexos/405000-409999/407429/norma.htm
- Central Bank of Nigeria, Payments System Policy Department, Regulations for the Operations of Non-Security Virtual Assets in Nigeria, November 2026, issued by circular PSP/DIR/PUB/CIR/001/004 of 5 November 2026, in force that day (circular, section 4.0); sections 37.11 to 37.13. https://www.cbn.gov.ng/Out/2026/CCD/REGULATIONS%20FOR%20THE%20OPERATIONS%20OF%20NON-SECURITY%20VIRTUAL%20ASSETS%20IN%20NIGERIA.pdf
- Bank of Ghana and Financial Intelligence Centre (Ghana), Anti-Money Laundering / Combating the Financing of Terrorism / Combating Proliferation Financing of Weapons of Mass Destruction (AML/CFT/CPF) Guideline for Accountable Institutions, September 2025, published by the Bank of Ghana on 12 December 2025, 2.4.2(3)(ii), 2.6.3(4)(vi) and 2.14. https://www.bog.gov.gh/wp-content/uploads/2025/12/BOG-FIC-AML-GUIDELINE-2025-FINAL-5-TO-BE-PUBLISHED.pdf
Ask us
A question about this rule?
A person at Archer reads every question and answers by email.
Got it. Where should we send the answer?
We’ll use your email to answer you. Privacy policy
Thanks. We’ve got your question.
The Fraud & AML Library is a practical reference, not legal advice. What binds is the regulator’s own text, which every rule and regulation page links.