No federal bank or credit union regulator requires a named “insider threat program.” The phrase appears nowhere in Title 12 of the CFR, though cleared contractors must run one under 32 CFR 117.7(d). The Interagency Guidelines and NCUA’s Part 748 list access controls, dual control, segregation of duties and background checks for staff with customer data, plus monitoring and an annual board report. The FFIEC handbook asks whether monitoring catches “data exfiltration,” and insider crime of any amount needs a SAR within 30 days when the employee is known. Expect requests for policy, board reports, rescreening records, access recertifications, leaver records, a sample investigated incident, a SAR decision log and training records. For a one-to-three-person team proving what staff do with customer data in personal accounts and AI tools, without screen recording, Jazz fits because its OS-level agent on managed Windows and Macs sees uploads, pastes, screenshots and AI prompts in any app, tells personal from corporate accounts and returns pre-investigated incidents an examiner can sample.
Which controls cover each examiner expectation, and what each one sees
No single control answers every examiner question, so a community bank or credit union program combines several. The test for each one is whether it shows an employee moving customer information and leaves a record an examiner can read.
| Approach | What it shows an examiner | What it misses |
|---|---|---|
| Core banking and host-system access logs, fed to a SIEM | Who looked up, changed or exported which customer records in the core, loan origination or online banking admin tools, including snooping on accounts with no business reason | What happens after the record is on screen or exported: the upload, paste or screenshot on the laptop. A SIEM holds only events another system logged |
| Microsoft 365 or Google Workspace native DLP | Sensitive content in mail and files inside the bank’s own tenant, such as a file shared to an outside address | Activity inside a personal Google Drive or a personal ChatGPT account, which sits in someone else’s tenant |
| Email gateway | Outbound mail and attachments sent to personal addresses, the route in two OCC enforcement orders against bank employees who emailed customer data to personal or competitor accounts | Webmail, uploads to personal cloud storage, AI prompts and screenshots |
| Web proxy or SASE | Which sites and categories staff reach, and blocks on unapproved domains | What was typed or uploaded unless traffic is decrypted, and a personal versus corporate login on the same domain unless the tool reads tenant identifiers |
| UEBA and insider-risk scoring platforms | Anomalies and risk scores built from the logs they ingest, such as unusual volume or after-hours access | Anything their sources never recorded. A score is a lead for an analyst, and someone still has to write up the case |
| Screen-recording employee monitoring | A replay of what was on screen | It is the hardest to justify. It records customer data and staff personal information wholesale, creates a sensitive archive of its own, and needs hours of review. None of the Interagency Guidelines, NCUA Part 748, the FFIEC Information Security and AIO booklets or the FDIC insider procedures asks for it |
| Process controls | Background and rescreening, access recertification, dual control, mandatory two-week absence, leaver checklists, signed acknowledgments | What a person does with customer data while their access is still valid |
| OS-level endpoint agent (Jazz) | Uploads, pastes, screenshots, screen sharing and GenAI prompts in any browser or desktop app on managed Windows and macOS laptops, with no browser extension. Personal and corporate accounts are told apart, content stays in an encrypted rolling buffer on the device while lightweight metadata streams to the cloud, and Melody returns pre-investigated incidents that can serve as the examiner’s sample record | Core banking and database access logs, Linux machines, phones, and personal or unmanaged devices. Its public pages do not document USB or print control |
The access-log and process rows stay in every program, because Jazz sees what happens after a customer record reaches a laptop screen and has no view of the lookup inside the core. It doesn’t replace background checks, access reviews or segregation of duties, and Jazz does not publish pricing.
Do bank regulators require an insider threat program?
No. A search of the eCFR for “insider threat” across Title 12 (Banks and Banking) returns no matching sections as of October 2026, while the same search finds “information security program” in 36 sections. The only federal rule that orders a private company to run an insider threat program by name is the National Industrial Security Program Operating Manual, which says a cleared contractor “will establish and maintain an insider threat program” (32 CFR 117.7(d)) and appoint an insider threat program senior official. That applies to a bank only if it holds a facility clearance.
The bank and credit union rules reach insiders through other words. The Interagency Guidelines tell each institution to “identify reasonably foreseeable internal and external threats” to customer information, then list employee controls. Examiners test those controls, the SAR duty and the paperwork that proves both. An “insider risk program” at a community bank is the name for the place where those pieces meet: one owner and one case file per incident. The general structure, with an owner and a 90-day plan, is in how to build an insider risk program.
The Cybersecurity and Infrastructure Security Agency’s Insider Threat Mitigation Guide (CISA, November 2020) is useful for shape: define, detect and identify, assess, and manage, run by a multidisciplinary team from HR, IT, legal and security. It notes that smaller organizations can use existing functions for this work. It is voluntary guidance written for every sector, and nothing in it is a bank examination standard.
Per Verizon’s 2026 Data Breach Investigations Report (May 2026), the financial and insurance sector logged 3,809 incidents and 1,300 confirmed data disclosures, internal actors were behind 12% of those breaches, down from 22% the year before, and Verizon says internal actor breaches “are mainly accidental.” Across all industries, Verizon describes the common convenience case as an employee who “emails company data to a personal account.”
What the regulations say about employees and customer information
The binding text is the Interagency Guidelines Establishing Information Security Standards, published three times with the same wording: 12 CFR Part 30, Appendix B for national banks and federal savings associations (OCC), 12 CFR Part 364, Appendix B for state nonmember banks (FDIC) and 12 CFR Part 208, Appendix D-2 for state member banks (Federal Reserve). A state-chartered bank also answers to its state supervisor.
Section II.A requires a written information security program “appropriate to the size and complexity” of the institution. Section III.C.1 then says the institution “must consider whether the following security measures are appropriate” and adopt those it concludes are. The employee-specific items are:
- III.C.1.a: access controls, including “controls to prevent employees from providing customer information to unauthorized individuals who may seek to obtain this information through fraudulent means.” The wording targets staff being tricked by outsiders, so deliberate exfiltration by an employee is reached through the risk assessment’s “internal” threats and the monitoring item.
- III.C.1.e: “Dual control procedures, segregation of duties, and employee background checks for employees with responsibilities for or access to customer information.”
- III.C.1.f: “Monitoring systems and procedures to detect actual and attempted attacks on or intrusions into customer information systems.”
- III.C.1.g: response programs, “including appropriate reports to regulatory and law enforcement agencies.”
- III.C.2: “Train staff to implement” the program, and III.C.3: regularly test key controls, with tests run or reviewed by people independent of those who maintain them.
Under III.A the board “shall” approve the written program, and under III.F management “shall report to its board” at least annually, covering “security breaches or violations and management’s responses.” That annual report is where an insider incident belongs.
Credit unions work from 12 CFR Part 748. Section 748.0 requires every federally insured credit union to keep a written security program that protects against “unauthorized access to or use of” member records. Appendix A repeats the bank language word for word with “member” in place of “customer,” including the dual control, segregation of duties and background check item, with one difference: its board approval and annual board report items say “should” where the bank appendices say “shall.” Appendix B adds that credit unions must run background checks to avoid hiring anyone barred by 12 U.S.C. 1785(d). Section 748.1(c) now requires a credit union to tell NCUA within 72 hours of a reportable cyber incident, which includes “a substantial loss of confidentiality” from “unauthorized access to or exposure of sensitive data.” A large insider data theft may meet that test even with no outage; that is a reading of the text for counsel to confirm.
Three other rules can apply:
- The OCC’s heightened standards (12 CFR Part 30, Appendix D) apply to banks with average total consolidated assets of $50 billion or more. The OCC proposed in December 2025 to raise that threshold to $700 billion. Both thresholds sit far above community-bank size.
- New York-chartered banks and credit unions are covered entities under the Department of Financial Services’ cybersecurity regulation. 23 NYCRR 500.14(a)(1) requires controls “designed to monitor the activity of authorized users and detect unauthorized access or use of, or tampering with, nonpublic information by such authorized users,” and 500.7 requires at least annual access reviews and prompt termination of access after departures.
- The federal 36-hour notification rule (12 CFR Part 53 and its FDIC and Fed twins) applies when an incident has materially disrupted or degraded, or is reasonably likely to materially disrupt or degrade, operations, a material business line or financial stability. A quiet data theft by one employee will usually fall outside it, though a bank that judges an incident reasonably likely to materially disrupt or degrade those operations must notify its regulator within 36 hours.
Amended Regulation S-P and the FTC Safeguards Rule bind broker-dealers, advisers and non-bank lenders rather than chartered banks; what those rules require, the FTC Safeguards checklist and the rules for staff AI use at banks are covered separately.
What examiners read: the FFIEC handbook, FDIC procedures and the OCC handbook
The FFIEC IT Examination Handbook’s Information Security booklet (September 2016) is written for banks and credit unions alike. Section II.C.7 lists the risks from internal users, among them “appropriation of strategic or customer data for espionage or fraud schemes” and “misuse of data following the termination or change in job responsibility of an employee.” Its subsections set the expectations an examiner tests:
- II.C.7(a): sensitive roles “may warrant additional screening and recurring background and credit checks.”
- II.C.7(b): least privilege, ongoing owner reviews of access, “timely notification from human resources to security administrators to adjust user access based on job changes, including terminations,” and periodic independent reviews.
- II.C.13(a): logging and monitoring over stored information, and “appropriate controls (such as the use of a DLP program) over portable devices.”
- II.C.13(e): policies telling employees they “are not authorized to use unsanctioned or unapproved IT resources (e.g., online storage services, unapproved mobile device applications, and unapproved devices).” A personal Google Drive or a personal AI account falls here.
- III.B: monitoring should cover “suspicious users, such as those who do not comply with or seek to evade security policies,” and outgoing traffic “seeking to identify malicious activity and data exfiltration,” with independent monitoring of administrators.
The booklet’s Appendix A examination procedures turn this into questions. Objective 2 asks whether the board receives an annual report covering breaches and management’s responses. Objective 6, procedure 8 asks whether management “effectively mitigates risks posed by users” through screening, a user access program, segregation of duties, confidentiality agreements and training. Procedure 18 asks for policies restricting unsanctioned online storage, and procedure 22 for periodic review of application access. Objective 8 asks for “monitoring both incoming and outgoing network traffic to identify malicious activity and data exfiltration” and for forensics capability. The appendix lets examiners use only the components that fit “the size, complexity, and nature of the institution’s business.”
The Architecture, Infrastructure, and Operations booklet (June 2021) adds two points for small teams. VI.A.4 says “background checks should be performed periodically during employment” and that in smaller entities “appropriately implemented rotation of duties can be an effective compensating control” for segregation of duties. III.B.3 says management “should employ appropriate data protection and data loss prevention tools to minimize the potential for exfiltration or misuse of sensitive customer and entity information.” That is handbook guidance on what examiners look for. No regulation requires a specific product.
The FDIC’s Insider Fraud and Abuse core analysis procedures (examination module dated October 2021) are the closest thing to an insider checklist. They ask whether policies cover “ongoing background screening to periodically review insider activity, credit reports (where allowed by law), employee email accounts, public records,” whether they restrict “unauthorized removal of documents, work papers, computer files, and other proprietary data and records from the institution’s premises by physical or electronic means, including customer personally identifiable information (PII),” and whether there is an insider fraud response plan. Background checks should include enforcement databases and criminal records. Examiners check for “an annual two-week absence from duties,” with remote access disabled during it, and banks without internal audit can point to “extensive separation of duties” and “strictly enforced mandatory vacation policies.” The module says examiners “are not expected to perform every procedure at every bank.” Section 9.1 of the FDIC’s Risk Management Manual, “Bank Fraud and Insider Abuse” (October 2025), is a list of lending and transaction red flags and says almost nothing about electronic customer data.
The OCC’s Insider Activities booklet (Version 1.0, November 2013) covers Regulation O lending, conflicts of interest and insider abuse. It says banks “may broadly define insiders to include all bank employees” and has examiners collect “whistle-blower procedures for confidential reporting of suspected fraud and insider abuse.” It does not address customer data leaving through employees, so that part sits in the FFIEC booklet and the FDIC module.
The FFIEC sunset its Cybersecurity Assessment Tool on August 31, 2025. It points institutions to NIST’s Cybersecurity Framework 2.0 and CISA’s Cybersecurity Performance Goals, says they may also consider the Cyber Risk Institute Profile and the CIS Controls, and states that these “are not examination programs.” NCUA continues to support its Automated Cybersecurity Evaluation Toolbox (ACET) for credit unions. A self-assessment against one of these is useful background, and examiners still test the controls themselves.
The examiner request list: an insider threat program template for banks and credit unions
Each request below traces to the Interagency Guidelines, the FFIEC Information Security and AIO booklets, the FDIC insider procedures, the OCC Insider Activities booklet or the SAR regulations named in its row.
| What the examiner asks for | What satisfies it | Evidence artifact to keep |
|---|---|---|
| A written program that covers insiders (Guidelines II.A and III.B.1; NCUA 748.0) | A board-approved information security program whose risk assessment names employee misuse of customer data, with a named owner for insider cases | The program, the risk assessment and the owner’s designation |
| Board oversight (Guidelines III.A and III.F; FFIEC Appendix A, Objective 2) | Board approval and an annual report that counts insider incidents under “security breaches or violations and management’s responses” | Board minutes and the annual report |
| Background checks and rescreening (Guidelines III.C.1.e; FFIEC II.C.7(a); AIO VI.A.4; FDIC procedures) | Pre-hire criminal and enforcement-database checks, recurring checks for sensitive roles, the same process for directors and officers | Screening records per person, the rescreening schedule and exceptions |
| Confidentiality, acceptable use and code of conduct (FFIEC II.C.7(d) and (e); FDIC procedures) | Signed agreements before access, annual acceptable-use acknowledgment, code of conduct signed by all institution-affiliated parties | Signed acknowledgments, dated |
| Rules on removing customer data and using unapproved tools (FDIC procedures; FFIEC II.C.13(e)) | A policy that restricts removal of customer PII by electronic means and names personal cloud storage, personal email and personal AI accounts, plus a list of approved AI tools | The policy with version history and the approved-tool list |
| Access recertification (FFIEC II.C.7(b) and II.C.15(b); Appendix A, Objective 6) | Periodic owner reviews of core and application access, HR notice to security on role changes and terminations | Signed recertifications and access-removal tickets |
| Dual control, segregation of duties, mandatory absence (Guidelines III.C.1.e; AIO VI.A.4; FDIC procedures) | A segregation-of-duties matrix, a two-week absence with remote access disabled, rotation of duties where staff are too few | The matrix, absence records and rotation schedule |
| Monitoring, and a sample investigated incident (Guidelines III.C.1.f; FFIEC III.B; Appendix A, Objective 8) | Coverage of core access logs and of data leaving laptops, plus one or two closed cases showing detection, investigation and decision | Case files with who, what data, from where, to which account, when, and the outcome |
| Leaver procedure (FFIEC II.C.7 and II.C.7(b)) | An activity review for every departing employee with customer-data access, access cut on the last day, devices returned | A per-leaver checklist with the review result and the access-termination time |
| SAR decision log (12 CFR 21.11, 353.3, 208.62, 748.1(d)) | A documented decision on every insider case, filed within 30 days when the employee is known, board notified of filings | Decision memos, SAR copies and supporting documents kept five years, board notices |
| Customer and regulator notice (Guidelines Supplement A; NCUA Appendix B) | A reasonable investigation into likely misuse, regulator notice as soon as possible for sensitive customer information, customer notice when misuse is reasonably possible | The investigation memo and the notice log |
| Training (Guidelines III.C.2; FFIEC II.C.7(e); FDIC procedures) | Annual training covering data loss through email and removable media, shadow IT and insider fraud awareness | Attendance records and the training content |
| Independent testing (Guidelines III.C.3; FDIC procedures) | Internal or external audit with insider threat in scope, or a documented rationale where there is no internal audit | Audit reports and remediation tracking |
| A way for staff to report concerns (OCC Insider Activities) | Confidential reporting without reprisal | The procedure and the complaint log |
The monitoring row needs two halves of evidence. Core logs prove who opened a record, and the examiner’s sample case also needs to show where the data went after it reached the laptop and whether the destination was a corporate or personal account. On managed Windows and Mac laptops, a Jazz investigation reconstructs that story across the data, the systems it moved between, the person and their role, and whether the movement fits a business process, which is the shape of a case file an examiner can read.
Do you have to file a SAR when an employee takes customer data?
Sometimes, and the decision itself is what examiners check. The SAR regulations of all four agencies carry a prong headed “insider abuse involving any amount”: 12 CFR 21.11(c)(1) for national banks, 12 CFR 353.3(a)(1) for FDIC-supervised institutions, 12 CFR 208.62(c)(1) for state member banks and 12 CFR 748.1(d)(1)(i) for credit unions. A filing is required when the institution detects a known or suspected federal criminal violation and has “a substantial basis for identifying one of its directors, officers, employees, agents or other institution-affiliated parties” as involved, “regardless of the amount involved.” The other prongs carry dollar thresholds: $5,000 where a suspect can be identified and $25,000 regardless of suspects.
The trigger is a suspected federal crime, and whether an employee’s data taking amounts to one is a call for counsel and the BSA officer. The rest is mechanical:
- File within 30 calendar days of initial detection. The extra 30 days, up to 60 in total, applies only when no suspect has been identified, so a known employee means 30 days.
- Management must promptly notify the board, or a designated committee, of each SAR filed.
- Keep the SAR and the “original or business record equivalent of any supporting documentation” for five years. Supporting documentation is deemed filed with the SAR.
- SARs are confidential. The HR and legal file on the employee should hold the facts, never a reference to whether a SAR exists.
FinCEN’s 2016 cyber-event advisory (FIN-2016-A005) and its FAQs are written around external cybercriminals and do not address employees as perpetrators, so the insider duty rests on the agency regulations. The response-program guidance attached to the Guidelines (Supplement A) runs alongside: notify the primary federal regulator “as soon as possible” after unauthorized access to sensitive customer information, investigate the likelihood of misuse, and notify customers when misuse “has occurred or is reasonably possible.” Sensitive customer information there means a name, address or phone number together with a Social Security number, account number, card number or account credentials.
What happens when bank employees take customer data: public cases
Public orders and filings show the pattern:
- BOKF, N.A. (Tulsa): An administrative assistant resigned on or about June 7, 2010 to join a competitor bank, then between June 10 and June 15 sent seven emails containing customer nonpublic personal information to her soon-to-be supervisor there. The OCC issued a personal cease-and-desist consent order on January 8, 2014 (OCC EA 2014-005), more than three years after the emails.
- Fidelity Bank of Florida, N.A.: A portfolio manager transferred customer files containing nonpublic personal information to his personal email accounts in January 2014. The OCC’s March 2015 consent order (EA 2015-025) imposed a $2,500 civil money penalty and found he caused the bank’s violation of its privacy rule, 12 CFR Part 40.
- Corning Credit Union: A financial adviser dually employed by the credit union and its broker-dealer partner downloaded data on more than 2,000 customers, including account numbers, balances and Social Security numbers, to a personal flash drive and laptop before leaving for a competitor in 2008. The SEC sustained a 90-day suspension in a December 11, 2014 opinion. The enforcement came through FINRA, not NCUA.
- Community Bank (CB Financial Services), Pennsylvania: The parent company filed an Item 1.05 Form 8-K on May 11, 2026 after the bank found an employee had handled non-public customer information, including names, Social Security numbers and dates of birth, with “an unauthorized artificial intelligence-based software application.” Executives later told American Banker (June 2026) that the employee used a personal account on a personal device even though the bank offered an approved AI tool.
The two OCC orders both involve email, which an email gateway can catch. The 2026 case involved an AI tool, a personal account and a personal device, which no control on bank-managed laptops, mail or tenant would have seen. The control for that gap is an access policy that keeps customer data off devices the bank doesn’t manage.
Worked example: a wealth adviser leaves for a competitor bank
The case below is hypothetical, built on the pattern in OCC enforcement orders against bank employees who sent customer data to personal email accounts or to a competitor bank. A wealth adviser at a state-chartered community bank gives two weeks’ notice to join a competitor bank. In the week before resigning, the adviser exports a client spreadsheet with names, account numbers and balances from the CRM, uploads it to a personal Google Drive, and pastes the balances of the largest relationships into a personal ChatGPT account to draft transition letters. All of it happens on a bank-managed laptop.
What the examiner will want to see, months later:
- The leaver record: when HR told security about the resignation, when the activity review ran, and the time access was cut.
- The investigated incident: which file, how many customers, which fields, which destination accounts, and when. Names with account numbers meet the definition of sensitive customer information, so this case starts the response-program steps.
- The response decisions: the regulator notice, the misuse investigation and the customer-notice decision, each dated.
- The SAR decision: counsel’s view on whether the facts amount to a federal crime with a substantial basis to name the adviser, the date of initial detection, the filing or the reasons for not filing, and the board notice if filed.
- The annual board report that lists the incident and the changes made afterwards.
What the bank can produce depends on its controls. With core logs and an email gateway alone, it can show the CRM export and an empty email record. It cannot show the Drive upload or the ChatGPT paste, so it cannot show whether the file left the bank or where it went, and the notice decision rests on guesswork. With an OS-level agent on the laptop, the bank has a case file that shows the export, the upload to a personal Google Drive account rather than the bank’s tenant, the paste into a personal ChatGPT account, the timeline and the adviser’s role. On managed Windows and Mac laptops, Jazz’s agent records those actions in any browser or desktop app, and the forensic data held in the encrypted buffer on the device can show what the file contained, which can narrow who gets a notice. Verizon’s 2026 report says an activity review triggered when someone with sensitive access resigns “has caught numerous breaches” in its dataset.
No laptop control can show what the adviser photographed with a phone or did on a personal computer. That gap belongs to access policy and to the confidentiality agreement the adviser signed.
Can a one-to-three-person security team meet these expectations?
Yes, if ownership is split and the evidence is kept as work happens. The Guidelines scale the program to “the size and complexity” of the institution, the FFIEC appendix lets examiners use only the components that fit, and the FDIC module accepts separation of duties and enforced vacations as mitigation where there is no internal audit. The AIO booklet names rotation of duties as a compensating control when there are too few staff to separate duties.
In practice HR owns screening, rescreening and the leaver trigger; operations owns access recertification and the segregation-of-duties matrix; the BSA officer owns the SAR decision log; and security owns monitoring, case files and the training record. The security team’s real constraint is analyst time, because a monitoring tool that produces thousands of raw alerts gives an examiner a backlog instead of a sample case. Jazz is built so a single analyst can run the whole program: in Jazz deployments, a typical 30 days produces about 2 million signals and around 80 incidents that reach an analyst, already investigated (Jazz, September 2026). More on running this with a small team is in handling insider threats as a one-person security team.
How Jazz fits a bank or credit union insider risk program
Jazz answers the laptop side of two request-list rows: data leaving laptops, and the sample investigated incident. Its endpoint agent sees copy and paste, screenshots, GenAI prompts, screen sharing and file uploads at the operating-system level across desktop apps, browsers, sanctioned SaaS, shadow IT and personal cloud accounts on managed Windows and macOS, without per-app integrations or a browser extension. Its Local Context Vault keeps rich forensic data in an encrypted, rolling buffer on the device, and the agent streams lightweight metadata about data movements to the Jazz cloud. A sanctioned corporate AI workspace and a personal login count as different destinations, so the same paste is judged differently in each.
Posture surfaces patterns such as repeated uploads to personal cloud accounts or large downloads before offboarding, which is the leaver review in the request list. Prevention responses run from asking the user for a justification, through escalation to a manager or analyst, to a targeted block, so a teller pasting account numbers into a personal chatbot can be stopped without blocking the approved tool. At UCLA Anderson, Jazz surfaced a user capturing screenshots of sensitive data, which the school’s IT security lead said no other DLP tool would have detected.
Cass Information Systems is a financial holding company supervised by the Federal Reserve, with a Missouri-chartered bank subsidiary, per its 2025 Form 10-K. Its CISO, Sean Mullins, says on Jazz’s homepage: “Meeting Federal Reserve expectations for data governance is critical. Jazz provides the deep forensic evidence and contextual analysis we need to prove we can protect our data in use.”
Jazz runs on managed Windows and macOS only, so Linux machines, phones and personal or unmanaged devices sit outside it, as the CB Financial upload did. It doesn’t read core banking or database access logs, so snooping inside the core stays with those logs and your SIEM. Its public pages do not document USB or print control. Background checks, access recertification, segregation of duties and the SAR decision remain people and process work, and Jazz supplies the record of what left a laptop and where it went.
Frequently asked questions
Do bank regulators require an insider threat program?
No federal bank or credit union regulator requires one by name. An eCFR search of Title 12 finds no section mentioning "insider threat"; cleared contractors face that duty under 32 CFR 117.7(d). Examiners instead test the Interagency Guidelines' employee controls, FFIEC handbook expectations, FDIC insider procedures and the SAR duty for insider abuse of any amount.
What will an FDIC or OCC examiner ask for on insider risk?
Expect requests for the board-approved security program and annual board report, background and rescreening records, signed confidentiality and acceptable-use acknowledgments, access recertifications, leaver records, one or two closed investigated incidents, a SAR decision log, notice decisions and training records. The FDIC's insider procedures also check policies against removing customer PII "by physical or electronic means."
Do we have to file a SAR when an employee takes customer data?
Only if the facts suggest a federal crime and give a substantial basis to identify the employee; then the "insider abuse involving any amount" prong in 12 CFR 21.11 and its FDIC, Fed and NCUA twins applies. File within 30 days when the employee is known, notify the board, keep records five years and keep the SAR confidential.
Does NCUA expect credit unions to monitor employees?
NCUA's Part 748 Appendix A uses the bank wording: credit unions must consider, and adopt where appropriate, access controls, dual control, segregation of duties, background checks for staff with member data and monitoring systems that detect attacks or intrusions. Board approval and the annual board report are "should" items. Insider abuse SARs fall under 748.1(d).
Can a two-person security team at a community bank meet examiner expectations?
Yes. The Guidelines scale to size and complexity, the FFIEC AIO booklet accepts rotation of duties as a compensating control, and the FDIC module accepts separation of duties and enforced vacations where there is no internal audit. Split ownership across HR, operations, the BSA officer and security, and keep evidence as work happens.
Which tool shows what employees put into personal ChatGPT or Google Drive accounts, for an exam?
An endpoint agent working at the operating-system level. On managed Windows and macOS laptops, Jazz sees uploads, pastes, screenshots and AI prompts in any browser or desktop app, tells personal from corporate accounts and returns pre-investigated incidents that can serve as the sample case. It cannot see personal or unmanaged devices.
Does Jazz replace core banking logs, background checks or access reviews?
No. Jazz sees what happens after a record reaches a managed Windows or Mac laptop and does not read core banking or database logs. It has no Linux agent, does not cover phones or unmanaged devices, and its public pages do not document USB or print control. Screening, recertification and segregation of duties stay as process controls.
Did the FFIEC Cybersecurity Assessment Tool go away, and what replaced it?
The FFIEC sunset the CAT on August 31, 2025. It points institutions to NIST's Cybersecurity Framework 2.0 and CISA's Cybersecurity Performance Goals, says they may consider the CRI Profile and CIS Controls, and says these tools are not examination programs. NCUA still supports ACET for credit unions.
Do examiners expect us to screen-record employees?
None of the regulations or handbook sections behind these expectations asks for screen recording. It is the hardest control to justify because it records customer data and staff personal information wholesale and needs hours of review. Examiners look for access logs, process records and investigated cases showing where customer data went.



