The FFIEC’s 2016 payment network warning treated a valid operator login as the attack path

On June 7, 2016, the FFIEC described attackers using valid operator credentials on trusted terminals. In our reading, that made privileged access, alerts on altered logging, provider notice and recovery the controls that mattered.

Looking Back revisits past weeks one at a time, up to the most recent, and asks what it meant for the banks, health systems and public agencies we serve. The series was written in 2026, with the benefit of hindsight, and each article says which week it looks back on.

Two-minute summaryThe dek and the first sentence of each section, quoted from the article

On June 7, 2016, the FFIEC described attackers using valid operator credentials on trusted terminals. In our reading, that made privileged access, alerts on altered logging, provider notice and recovery the controls that mattered.

  1. The week The document cited recent cyber attacks.
  2. The credential: a valid login as the attack path Payment operations rest on a familiar control.
  3. The terminal and its record: noticing a changed environment The second part concerns the environment where messages are created.
  4. The provider: what the bank needs from third parties The third part is the boundary with suppliers.
  5. The money: recovery and the payment that has left The fourth part begins where prevention ends.
  6. What the statement left to each bank The statement was short, and much of its force came from what it chose not to settle.
  7. Objections to this reading Three objections deserve a direct answer.
  8. Checks a technology or risk leader can run now These checks follow from the statement’s structure.
  9. With hindsight The 2016 statements named no institution.

The week

On June 7, 2016, the Federal Financial Institutions Examination Council issued a joint statement titled Cybersecurity of Interbank Messaging and Wholesale Payment Networks. [1] [2] The council’s member bodies include the three federal banking agencies and the NCUA, which supervises credit unions. [1]

The document cited recent cyber attacks. Its stated purpose was to remind financial institutions to actively manage the risks of interbank messaging and wholesale payment networks. [1] It asked institutions to review their controls over authentication, authorization, fraud detection and response management. [1]

The council was explicit about the document’s weight: it added no new regulatory expectations, in its own words. [1] It was meant to alert institutions to risk mitigation techniques for attacks that exploit vulnerabilities and gain unauthorized entry through trusted client terminals running messaging and payment networks. [1]

The background section described what recent attacks had shown attackers could do. The list had five items. [1]

  1. Compromise an institution’s wholesale payment origination environment, bypassing information security controls. [1]
  2. Obtain and use valid operator credentials with authority to create, approve and submit messages. [1]
  3. Employ a sophisticated understanding of funds transfer operations and operational controls. [1]
  4. Use highly customized malware to disable security logging and reporting, and other operational controls, to conceal and delay detection. [1]
  5. Transfer stolen funds across multiple jurisdictions quickly to avoid recovery. [1]

The document named two kinds of risk. Unauthorized transactions could expose the originating bank to financial loss and to compliance risk. [1] A footnote gave the USA PATRIOT Act, the Bank Secrecy Act and the Office of Foreign Assets Control as examples. [1]

It then set out steps an institution should consider. These were: ongoing information security risk assessments; security monitoring, prevention and risk mitigation; protection against unauthorized access; regular testing of controls around critical systems; business continuity; awareness training; and participation in information-sharing forums. [1]

For regulatory expectations it pointed to existing examination booklets, among them the one on wholesale payment systems. [1] It also told institutions to consult their payment system provider’s guidance for specific control recommendations. [1]

Its further reading included two customer communications from the SWIFT messaging network, dated May 13 and May 20, 2016. [1] It also listed the Federal Reserve Banks’ Operating Circular No. 5 on electronic access, effective June 30, 2016. [1]

Two member agencies issued it to their own institutions the same day. The OCC issued it as Bulletin 2016-18, dated June 7, 2016. [2] The bulletin said the guidance applied to all OCC-supervised institutions that use interbank messaging and wholesale payment networks, either directly or through service providers. [2]

The FDIC issued it as Financial Institution Letter FIL-37-2016, also dated June 7, 2016, and said the letter applied to all FDIC-supervised institutions. [3] Its highlights said recent attacks had resulted in large-dollar fraud at several foreign institutions. [3] The FDIC suggested routing the letter to the chief executive officer, the chief information officer and the chief information security officer. [3]

None of the three documents named an institution that had been attacked. They described the attack pattern and left the identification to others. [1] [2] [3]

The June 2016 statement can be read as a reminder list. It said so itself: no new expectations, a pointer to existing handbook booklets, and a set of familiar controls. [1] Most of its steps would fit any information security guidance.

Our argument is that its background section changed the problem the controls had to solve. It said attackers had obtained and used valid operator credentials with the authority to create, approve and submit messages. [1] A payment message created with a valid credential, on the bank’s own terminal, passes every check that asks whether the sender is authorized. The statement treated that kind of message as the threat.

Read that way, the statement asked a bank to stop treating the trusted terminal as the edge of its defenses. It asked for layers that would still work after an attacker held a real login. In our reading, that is why the steps that matter most are about privilege, logging, providers and recovery, not about the perimeter.

We take the statement in four parts, each a different place where a bank could stop a message sent with stolen authority:

  1. The credential: who can create, approve and send a payment message, and how far a stolen login can reach.
  2. The terminal and its record: how the bank would notice that the environment itself had been changed, including its logs.
  3. The provider: what the bank needs from the third parties that run, host or connect its payment systems.
  4. The money: how fast the bank can recover operations and act on a payment that has already left.

The parts do not overlap. The first is about identity, the second about the integrity of the environment, the third about the boundary with suppliers, and the fourth about what happens after a message is sent. Together they cover the path of a fraudulent message from login to settlement.

After the four parts we set out what the statement left to each bank, the strongest objections to this reading, and a set of checks a technology or risk leader could run today.

The credential: a valid login as the attack path

Payment operations rest on a familiar control. One person creates a message, another approves it, and the system releases it only when both have acted. The design assumes that two credentials mean two people.

The statement’s background list breaks that assumption. It described attackers who obtained valid operator credentials with authority to create, approve and submit messages. [1] If an attacker controls both the maker’s and the checker’s credentials, dual control becomes one actor using two accounts.

The statement’s access steps read as an answer to that problem. It asked institutions to limit the number of credentials with elevated privileges, especially administrator accounts. [1] It asked them to limit how easily elevated privileges to critical systems could be assigned. [1]

Privilege and its review

The statement asked for periodic reviews of access rights to confirm approvals were still appropriate to the job. [1] It asked for regular audits of the access and permission levels of employees and contractors on critical systems. [1] It asked for least-privilege policies across the enterprise and said users should not have local administrator rights on workstations. [1]

Each of those steps shrinks the set of accounts an attacker can usefully steal. A payment operator without local administrator rights is a less useful foothold. An administrator account that cannot also approve payments is a less useful prize.

In our reading, the steps also implied a separation the statement did not spell out: the people who administer the payment platform should not be the people who approve payments on it. A credential that can change the system’s parameters and also release a message defeats the design.

Dormant and default credentials

The statement asked for stringent expiration periods for unused credentials. It asked banks to monitor logs for the use of old credentials and to terminate unused or unwarranted ones promptly. [1] It also asked them to change default passwords and settings for system-based credentials. [1]

Those steps target accounts nobody watches. A departed operator’s login, a service account created at installation, or a vendor account left active after a project are all credentials with authority and no daily owner. The statement treated their use as a signal worth logging.

Conditions on the login itself

The statement asked institutions to set authentication rules such as time-of-day and geolocation controls, or to use multifactor authentication for web-based control panels. [1] It asked for limits on sign-on attempts to critical systems, with accounts locked once the threshold was exceeded. [1]

Time-of-day rules matter for payments. A wire release at an hour when the operations team is not working is unusual, and a rule can stop it or route it for review. The statement offered such rules as an alternative to multifactor authentication, not as a requirement. [1]

Multifactor authentication has a limit the statement’s background list makes clear. If the attacker controls the terminal where the operator signs in, a second factor protects the login but may not protect the session that follows. That is a reason the statement asked for more than one layer. [1]

How the credential problem differs by institution

A large bank with its own messaging interface has many operators, several approval tiers and its own administrators. Its problem is the number of privileged accounts and the speed of access reviews.

A community bank or credit union may have two or three people who can release wires. One of them may also administer the workstation. For that institution, separating administration from approval may mean a second person, a provider-managed workstation, or a callback step, each with a cost.

A bank that reaches the networks through a correspondent or service bureau has a different question. Its operators’ credentials sit in a provider’s portal. The OCC applied the guidance to institutions using the networks directly or through service providers, so the indirect route did not take the bank out of scope. [2]

The evidence a credential program leaves

A control an institution cannot show is hard to defend in an examination. For credentials, the statement’s steps leave a natural trail: an inventory of privileged accounts, records of periodic access reviews, and logs of terminated and expired credentials. [1]

In our reading, the most useful single record is a list of every identity that can move money, with its owner, its role and the date its access was last confirmed. It answers the statement’s first question directly: how many credentials with elevated privileges exist, and who approved them. [1] Without that list, a bank cannot tell how much a stolen login is worth.

A second record is the configuration of approval rules in the payment system itself: who can approve what, up to what amount, and whether the system blocks one user from approving their own message. Payment platforms store those rules as settings. Settings can be changed, which is why the statement’s alert on changed baseline controls matters here too. [1]

The trade-offs of tighter credentials

Each credential step has a cost. Fewer privileged accounts mean fewer people who can fix a problem at night. Short expiration periods mean more resets and more help desk calls, and the help desk becomes a route attackers may try.

Time-of-day rules collide with real business. A corporate customer may need a late wire before a deadline, and a rule that blocks it creates pressure to grant an exception. A bank that adopts such rules needs a documented exception path that is itself controlled. Otherwise the exception becomes the weakness.

Small institutions feel the staffing cost most. Dual control needs two qualified people available whenever payments must go out, including during vacations and illness. A bank with three wire operators may find that strict separation leaves no cover. The statement did not address staffing, and it did not lower its expectations for small institutions. [2] [3]

Vendors and contractors with payment access

The statement named contractors alongside employees in its access audits. [1] Payment platforms can be installed, upgraded and supported by outside staff, some of whom may hold administrator rights during the work. Those accounts can outlast the project that created them.

A bank can treat each vendor session as a planned event: opened for a named task, watched while open, and closed when the task ends. That approach is our suggestion. It answers the statement’s concern about unused and unwarranted credentials for the accounts most likely to be forgotten. [1]

The terminal and its record: noticing a changed environment

The second part concerns the environment where messages are created. The statement said attackers had compromised the payment origination environment and bypassed information security controls. [1] It also said they had used customized malware to disable security logging and reporting. [1]

That last item matters more than it first appears. Most detection depends on logs. If an attacker can switch off the logs on the system that sends payments, the bank’s monitoring sees silence rather than an alarm.

Baselines and alerts on change

The statement asked institutions to establish a baseline environment so that anomalous behavior could be detected. [1] It asked them to implement alert systems to notify employees when baseline controls are changed on critical systems. [1]

In our reading, an alert on a changed control is the answer to malware that disables logging. The question it asks is not whether a payment looked odd but whether the system that produced it had been altered. A logging service that stops on the payment server should itself be an event someone sees.

Those alerts need to go somewhere the attacker does not control. If they are generated and stored on the same host that the malware has compromised, they can be suppressed as easily as the logs. A separate collector, with its own administrators, keeps the record outside the attacker’s reach. That is our inference, not text from the statement.

Monitoring and vulnerability management

The monitoring steps were broad. The statement asked banks to keep intrusion detection and antivirus protection up to date, configure firewall rules properly and review them periodically, and monitor alerts to identify, prevent and contain attacks. [1]

It asked for software assurance practices for internally developed applications, due diligence of third-party software and services, and penetration testing and vulnerability scans as necessary. [1] It asked banks to manage vulnerabilities promptly based on risk, track mitigation, and patch all applications, services and systems. [1]

It also asked them to prevent unpatched systems, such as home computers and personal mobile devices, from connecting to internal-facing systems. [1] And it asked for monitoring controls to detect unauthorized devices on internal networks. [1]

Limiting what the terminal can reach

Two steps aimed at narrowing the terminal’s exposure. The statement suggested filtering internet access through website whitelisting where appropriate, so employees reach only the sites their jobs require. [1] It asked for secure connections, such as virtual private networks, when systems are accessed remotely. [1]

A payment workstation that can browse the open web is a target for phishing and malicious downloads. A workstation that can reach only the payment interface and a few named services has a smaller attack surface. The statement did not require a dedicated workstation, but its steps point in that direction, in our reading.

Testing and reporting to the board

The statement asked institutions to test the effectiveness and adequacy of controls periodically. [1] It asked them to report the results to senior management and, if appropriate, to the board or a board committee, with recommended mitigation and progress on remediation. [1]

That reporting line turns a technical question into a governance record. An examiner reviewing a bank after the statement could reasonably ask whether payment system controls had been tested and whether the board had seen the results. Whether a particular examiner did so is not something the statement tells us.

Data handling on the payment path

The testing step carried several sub-items. The statement asked banks to encrypt sensitive data in transit and, where appropriate, at rest, on internal and external systems. [1] It asked for an adequate password policy, a review of password recovery processes, and procedures for destroying media based on the sensitivity of the information. [1] It asked for incremental and full backups of important files, stored offline. [1]

Password recovery deserves attention in a payments context. If an attacker can reset an approver’s password through a help desk or a self-service flow, the approval control is only as strong as that flow. The statement named the process as something to review. [1]

Why the record matters after the fact

Logs serve detection, and they also serve the investigation that follows. After a fraudulent payment, a bank needs to reconstruct which messages were created, by which credentials, from which machines, and what else the attacker touched.

The statement said the malware it described was used to conceal and delay detection of fraudulent transactions. [1] If the bank’s only copy of its payment logs sat on the compromised systems, the investigation may start with gaps. A copy held elsewhere, and kept long enough, lets the bank answer the questions customers, insurers and regulators will ask.

How long to keep payment system logs is a decision the statement left open. A bank should set it with its legal and records teams, taking account of its own record-keeping obligations.

Who watches the alerts

A large bank may run its own security operations center, so an alert on a changed control reaches its own analysts. A community bank may rely on its core provider or a managed security service for monitoring.

In that case the alert reaches the provider first. The bank’s protection then depends on what the provider is contracted to watch, how fast it must tell the bank, and whether the payment workstations are in scope at all. That links the second part of this analysis to the third. The statement’s contract clause on incident reports is where the two meet. [1]

Reconciliation as a second view

The statement did not use the word reconciliation. In our reading, it is the control that best fits its warning about concealment. If malware can alter what the bank’s own systems display, the bank needs a second view of what was sent.

That second view can come from outside the compromised environment: the network’s or correspondent’s record of messages and settlements, compared with what the bank’s systems say. A mismatch is a signal the attacker cannot easily hide, because it is produced elsewhere. How each network provides such records is a matter for its own guidance, which the statement told banks to follow. [1]

The provider: what the bank needs from third parties

The third part is the boundary with suppliers. We would expect few payment systems to be built and run entirely in-house. Messaging interfaces, core banking platforms, hosting and network connections may come from others.

The statement addressed this inside its risk assessment step. It asked institutions to ensure that third-party service providers perform effective risk management and implement appropriate controls. [1] It asked that providers properly maintain and regularly test their security controls, simulating potential risk scenarios. [1]

It also asked that providers be contractually obligated to provide security incident reports when issues arise that may affect the institution. [1] In our reading, that was the most concrete demand in the document. It named a contract term, not a principle.

Why a contract clause

A bank can only act on an incident it knows about. If a provider’s environment is compromised and the provider has no duty to say so, the bank may learn of it only when a payment goes wrong.

A contract clause turns notice into an obligation with a remedy. It also forces specifics: what counts as an incident, how fast notice must come, to whom, and with what detail. The statement did not set those terms. It left them to each bank and its provider. [1]

Services the bank provides to others

The statement asked institutions to review their risk management practices, including services provided to clients. [1] A bank that gives correspondent customers access to payment services is a provider to them. Its controls protect their payments as well as its own.

For a correspondent bank, the statement therefore had two faces. It was a customer of its own suppliers and a supplier to its respondent banks. The same contract and notice questions ran in both directions, in our reading.

The payment system provider’s guidance

The statement told institutions to consult their payment system provider’s guidance for specific security control recommendations. [1] It told them to review and adhere to technical guidance issued by payments and settlement networks for managing risks to critical systems. [1]

That instruction made the network’s own guidance part of the reference set. The statement listed two SWIFT customer communications from May 2016 and the Federal Reserve Banks’ Operating Circular No. 5 among its resources. [1] A bank using more than one network would need to track each one’s guidance and map it to its own controls.

Testing with providers

The continuity step asked institutions to coordinate business continuity development and testing with all applicable third parties, and to coordinate testing with other industry players. [1] A recovery plan that assumes a provider will be available, without having tested it with that provider, is an assumption rather than a plan.

For a community bank or credit union whose core, internet banking and wire platforms come from one provider, coordinated testing may be the only realistic route. The provider runs the infrastructure. The institution can test only what the provider lets it test.

What to ask a provider before signing

The statement gave the outcome a bank should seek: providers that manage risk, test their controls and are bound to report incidents. [1] Turning that into due diligence means asking specific questions. These are our suggestions, not text from the statement.

  • Which of the provider’s staff can reach the bank’s payment environment, and how is that access granted, reviewed and logged?
  • How does the provider test its controls, and will it share the results or a summary?
  • What does the provider count as an incident that may affect the bank, and how quickly will it say so?
  • Which other firms does the provider rely on to deliver the service, and does the notice duty pass through to them?

Concentration across institutions

Smaller banks and credit unions may buy core processing and payment services from the same providers. An attack on one provider’s environment could reach many institutions at once.

The statement’s instruction to coordinate continuity testing with third parties and with other industry players fits that situation. [1] A single institution cannot test a provider’s recovery alone. Testing organized by the provider, or through an industry group, may be the only test that covers the shared environment.

Credit unions and their service organizations

Credit unions fall within the statement through the NCUA, one of the FFIEC members on whose behalf it was issued. [1] Some may reach payment networks through corporate credit unions or credit union service organizations. For them, the provider questions above apply to those organizations as much as to any technology vendor.

The money: recovery and the payment that has left

The fourth part begins where prevention ends. The statement said attackers could transfer stolen funds across multiple jurisdictions quickly to avoid recovery. [1] Once a fraudulent message is released, the bank’s problem becomes speed: how quickly it notices and how quickly it can ask others to stop or return funds.

Business continuity for payment processing

The statement asked institutions to validate that business continuity planning supports the ability to quickly recover and maintain payment processing. [1] After a compromise, a bank may need to take its payment environment offline to investigate. It still has payments to make for customers.

A bank with a fallback route, such as a separate channel through a correspondent or a manual process with call-back verification, can keep paying while it investigates. A bank without one faces a choice between continuing on a suspect system and stopping customer payments. The statement asked banks to plan for that choice in advance, in our reading. [1]

Compliance risk after the fact

The statement’s risk section linked fraud to compliance. Unauthorized transactions could expose the originating bank to compliance risk, with the PATRIOT Act, the Bank Secrecy Act and OFAC named as examples. [1]

That footnote has practical weight. A fraudulent payment is a theft and may also be a transaction the bank must assess for suspicious activity reporting and sanctions screening. The fraud team, the BSA officer and the sanctions team each have a role. Which filings apply depends on the facts, and a bank should confirm them with counsel.

Information sharing as part of recovery

The last step asked institutions to take part in information-sharing forums. The statement named the Financial Services Information Sharing and Analysis Center and the government’s US-CERT as resources. [1] It said that because threats and tactics change rapidly, sharing can improve an institution’s ability to identify attack tactics and mitigate attacks. [1]

Sharing works in two directions during an incident. A bank that has been attacked can warn others about the method. A bank that has not been attacked can check whether the published indicators appear in its own logs, if it still has those logs.

Training that reflects the job

The statement asked for regular, mandatory security awareness training, including how to identify and prevent phishing. [1] It asked that training reflect the functions employees perform. [1] For payment staff, that means training on the specific ways a payment can be manipulated, beyond general phishing awareness.

Payment operators are the people most likely to notice something odd: a message they did not create, a failed reconciliation, or a missing confirmation. Training them to report those signals quickly is part of recovery as much as prevention.

The first hours after a suspect payment

The statement did not set out an incident playbook for payments. Its steps imply one, in our reading, and the order matters because funds can move across jurisdictions quickly. [1]

First, someone must be able to decide quickly whether to stop sending. That person needs authority written down in advance, because stopping payments affects customers. Second, the bank needs to contact its correspondent or network to ask about recalling or holding the payment. The contact details and the procedure should be known before they are needed.

Third, the bank must preserve evidence. Rebuilding a workstation restores service but can destroy the record of what the attacker did. Fourth, the bank must decide whom to tell: its board, its regulator, law enforcement and its customers. In 2016 the statement pointed to information-sharing forums and US-CERT, and set no regulator notice duty. [1]

Those four steps pull in different directions. Speed favors stopping and rebuilding. Evidence favors preserving. Customers favor resuming. A plan that names who decides each trade-off is worth more than a plan that lists steps without owners.

Insurance and the proof of loss

A bank that suffers a payment fraud may look to its insurance. Policies vary, and the statement said nothing about insurance. What it did say bears on the claim indirectly: a bank that can show its controls, its logs and its provider notices is better placed to document what happened and when. [1] Whether a given loss is covered is a question for the policy and for counsel.

Talking to customers

If a bank pauses its payment environment, corporate customers waiting on wires will notice. The statement’s continuity step asked banks to be able to maintain payment processing. [1] Where that fails, the bank needs a way to tell customers what is delayed and what alternative it offers, without revealing details that would help the attacker. That message is easier to write before an incident than during one.

What the statement left to each bank

The statement was short, and much of its force came from what it chose not to settle. Four gaps stand out, and each left a decision with the institution.

No notice clock. It asked for contractual incident reports from providers but set no time limit, no definition of an incident and no content. [1] Each bank negotiated its own terms, and smaller banks had less bargaining power in that negotiation, in our reading.

No definition of critical systems. It asked for controls around critical systems but did not list them. [1] A bank had to decide whether its payment workstation, its core banking platform, its directory service and its email were all critical. An attacker who controls the directory can create payment credentials, so the boundary matters.

No mandated technique. It offered multifactor authentication as one option for web-based control panels, alongside time-of-day and geolocation rules. [1] It did not require any single technology. That left room for judgment and also for inconsistency between institutions.

No new reporting duty to regulators. The statement asked banks to share information through forums such as FS-ISAC. [1] It did not create a duty to notify the agencies of a payment system compromise. That came later, under a separate rule we discuss in the hindsight section. [5]

Each of these gaps was consistent with the statement’s claim that it set no new expectations. Together they meant two banks could both follow the statement and end up with very different defenses. [1]

Objections to this reading

Three objections deserve a direct answer.

  1. The statement said it was not new. The document says so in plain words. [1] Our argument is not that it created new rules. It is that its background section described a specific attack path, valid credentials on a trusted terminal, and that the listed controls make most sense as layers against that path. A reminder can still change what a bank looks at first.
  2. The statement’s steps were generic. Many of them were. Patching, firewalls and training appear in most security guidance. [1] The steps that were specific to the attack, such as alerts on changed baseline controls, limits on elevated privileges and contract terms for incident reports, are the ones this reading puts first.
  3. The networks themselves would handle it. The statement told banks to follow their payment system provider’s guidance. [1] It also kept the bank responsible for its own environment and its own providers. The network can set standards for the interface. It cannot see a bank’s workstation or its logs.

A fourth point limits the argument. The statement addressed institutions supervised by the FFIEC member agencies. [1] Broker-dealers and investment advisers are supervised by other regulators, and the statement did not address them. Their payment controls answer to different rules.

Checks a technology or risk leader can run now

These checks follow from the statement’s structure. They are good practice that follows from it, not requirements stated in this form, and each institution’s obligations should be confirmed with counsel and its examiners.

Credentials

  • List every account that can create, approve or release a payment message, including service and vendor accounts, and confirm each has a named owner.
  • Confirm that no account can both administer the payment platform and approve a payment.
  • Confirm that unused payment credentials expire and that use of an expired or dormant credential raises an alert.
  • Confirm how a payment approver’s password can be reset, and who can do it.

Environment

  • Confirm that stopping or changing logging on payment systems generates an alert outside those systems.
  • Confirm that the payment workstations’ internet access is limited to what the job requires.
  • Confirm when payment system controls were last tested and when the board or its committee last saw the results.

Providers

  • Read the incident notification clause in each payment-related provider contract and record what it requires, by when and to whom.
  • Confirm whether business continuity tests for payments have included the providers that run the payment path.
  • For correspondent services the bank offers, confirm what it has promised its own customers about incident notice.

Recovery

  • Confirm there is a documented fallback way to make urgent payments if the main payment environment is taken offline.
  • Confirm that fraud, BSA and sanctions staff are named in the payment incident plan.
  • Confirm that logs from payment systems are kept long enough, and outside the systems themselves, to check shared indicators after the fact.

With hindsight

The 2016 statements named no institution. A federal charging document later described an attack of the kind they warned about. On September 6, 2018, the Justice Department announced charges in a criminal complaint filed on June 8, 2018, in Los Angeles federal court. [4]

The department said the complaint alleged that a conspiracy linked to the North Korean government stole $81 million from Bangladesh Bank in February 2016. [4] According to the allegations, the conspiracy compromised the bank’s network with spear-phishing emails and accessed its computer terminals that interfaced with SWIFT. [4] It then sent fraudulently authenticated SWIFT messages directing the Federal Reserve Bank of New York to transfer funds to accounts in other Asian countries. [4]

The release said the conspiracy attempted to and did gain access to several other banks in various countries from 2015 through 2018, using similar methods and watering hole attacks. [4] It said those operations attempted the theft of at least $1 billion. [4] The department said the charges were merely accusations and that the defendant was presumed innocent unless proven guilty. [4]

Those allegations match the pattern the FFIEC described in June 2016: a compromised origination environment and messages that looked authentic. [1] [4] They do not tell us whether any U.S. bank was attacked the same way, and we do not claim that.

A duty close to the statement’s contract clause for provider incident reports later became regulation, set by rule rather than left to each contract. On November 23, 2021, the OCC, the Federal Reserve Board and the FDIC published a final rule on computer-security incident notification. [5] It required a banking organization to notify its primary federal regulator of a notification incident as soon as possible and no later than 36 hours after determining one had occurred. [5]

The same rule set a separate duty for bank service providers. A provider had to notify each affected banking organization customer as soon as possible after determining it had a qualifying computer-security incident. [5] The test was an incident that caused, or was reasonably likely to cause, a material service disruption or degradation for four or more hours. [5] The rule was effective April 1, 2022, with a compliance date of May 1, 2022. [5]

Credit unions followed under a separate rule. On March 1, 2023, the NCUA published a final rule on cyber incident notification. [6] A federally insured credit union must report a reportable cyber incident as soon as possible and no later than 72 hours after it reasonably believes one has occurred. [6] The definition reached incidents caused by a compromise of a credit union service organization, cloud provider, managed service provider or other third-party data hosting provider, or by a supply chain compromise. [6] The rule was effective September 1, 2023. [6]

Third-party guidance moved too. On June 9, 2023, the Board, the FDIC and the OCC published final interagency guidance on third-party relationships, final as of June 6, 2023, replacing each agency’s existing general guidance. [7] It said reviewing a third party’s incident reporting and management processes helps confirm that its escalation and notification meet the bank’s expectations and regulatory requirements. [7]

That guidance is now under review. On September 15, 2026, the OCC, the Board, the FDIC and the NCUA proposed new third-party risk management guidance. [8] The agencies said the 2023 guidance had frequently been interpreted in an overly broad manner, with insufficient focus on tailoring. [8] They said any finalized guidance would replace the 2023 guidance and its supplemental resources. [8]

The proposal still pointed to incident notice. It said banking organizations and bank service providers have notification obligations under the agencies’ computer-security incident notification rule, which may inform ongoing monitoring of third parties. [8] As checked on 2026-11-16, the document we read is a proposal whose comment period closes that day, and we found no final version. [8]

So what held up? The 2016 statement’s most specific demand, that providers tell banks about incidents, is now partly a federal rule: bank service providers must notify affected banking organization customers of incidents that disrupt, or are reasonably likely to disrupt, services for four or more hours. [5] Its view that a bank answers for its own environment and its providers is still in the guidance, though the 2026 proposal would scale the effort to assessed risk. [7] [8]

The later rules dealt with notice and oversight, not with the attack path itself, in our reading. The 2018 allegations describe messages that were fraudulently authenticated, the kind no network check would reject on its own. [4] In our view, the controls that address that path are still the ones the statement listed: fewer privileged credentials, alerts when controls change, provider notice and a way to keep paying while investigating. [1]

We have not reviewed examination findings on how banks applied the statement, and none appear in the sources we checked. Institutions should confirm their current obligations with counsel.

Spotted an error? Tell us and we will check it and correct the article where needed.

  • FFIEC
  • wholesale payments
  • SWIFT
  • privileged access
  • third-party risk
  • incident notification