SEC examiners’ 2015 cyber request list sought dates and records showing access controls were in use

A September 2015 SEC staff risk alert listed what examiners might request, including the date each former employee lost access, systems without multifactor authentication and incident discovery dates. In a breach order issued September 22, destroyed logs left the scope unknown.

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

A September 2015 SEC staff risk alert listed what examiners might request, including the date each former employee lost access, systems without multifactor authentication and incident discovery dates. In a breach order issued September 22, destroyed logs left the scope unknown.

  1. The week It gave a reason.
  2. The access record: dates that prove a policy ran The appendix asked for firm policies on establishing employee access rights, including role or group membership, and on updating or terminating access based on personnel or system changes. [1] That part could be answered from a policy manual.
  3. The data record: knowing where customer information lives It also asked about the systems, utilities and tools used to prevent, detect and monitor data loss relating to personally identifiable information and access to customer accounts. [1] The order of the requests is telling.
  4. The vendor record: third parties inside the perimeter In our reading, that second category is easy to overlook.
  5. The incident record: plans, tests and the events themselves It asked about the firm’s process for testing or exercising its incident response plan, including frequency, reports and any remediation. [1] A plan that had never been exercised would produce an empty answer to that request.
  6. Who the alert reached, and who it did not For a large broker-dealer with dedicated security staff, the appendix was a checklist against existing evidence.
  7. Objections to this reading The alert said the Commission had expressed no view, that it was not legal advice and that it did not alter existing requirements. [1] A firm could reasonably say it created no new obligation.
  8. Checks a CISO or chief compliance officer could run These checks are our suggestions, drawn from the 2015 appendix.
  9. With hindsight On August 7, 2017, OCIE published its observations from the 2015 initiative.

The week

On September 15, 2015, the SEC’s examination office published a risk alert announcing the focus of its 2015 cybersecurity initiative. [1] The document described a second round of reviews of registered securities firms, and attached a sample list of the information staff might request. [1]

The document traced the work back to a roundtable the SEC sponsored in March 2014, a first round announced in April 2014, and a summary published in February 2015. [1] It said this round would involve more testing to assess whether firm procedures and controls were implemented. [1]

It gave a reason. It said public reports had identified breaches related to weaknesses in basic controls, and cited a February 2015 FINRA report on cybersecurity practices. [1] A footnote recalled that Regulation S-P requires the registered firms it covers to adopt written safeguards for customer information. [1]

The main text named basic controls whose absence may put a firm at risk of a breach. Its examples were multifactor authentication and updating access rights based on personnel or system changes. [1] It said some breaches may have resulted from weaknesses in controls over patch management and system configuration. [1]

It said staff may look at how a firm monitors the volume of content its employees or third parties send outside the firm, such as by email attachments or uploads. [1] It said they may review how a firm verifies that a customer’s request to transfer funds is authentic. [1]

On third parties, it said some of the largest data breaches of recent years may have resulted from the hacking of vendor platforms. [1] It said employees and vendors could be a firm’s first line of defense with proper training, by alerting IT staff to suspicious activity. [1]

The appendix was the more specific part. It said some of its questions tracked the NIST Framework for Improving Critical Infrastructure Cybersecurity, released on February 12, 2014. [1] It said the list was not all-inclusive and that requests would vary with each firm’s business model, systems and technology environment. [1]

Its requests asked for records as well as policies. It asked for former employees’ last date of employment beside the date their system access ended, and for reassigned employees’ move dates beside the date their access changed. [1] It asked which systems used multifactor authentication and which did not. [1]

It asked for instances of access granted contrary to policy, for data maps showing ownership of personal information, and for documentation of monitoring for exfiltration by email, physical media or web transfer. [1] It asked for vendor notices of system changes and for contingency plans covering a vendor’s bankruptcy. [1]

On incidents, it asked for test reports for the response plan, and for each incident’s date, discovery process and escalation. [1] It asked for customer losses, amounts reimbursed and cyber insurance claims and recoveries. [1]

The document carried the usual limits. It said it was not legal advice, that its factors gave no safe harbor, and that adequacy could be judged only against each firm’s own profile. [1]

The safeguards rule the document cited soon reached an enforcement order against an investment adviser, which we take up in the hindsight section, since it falls after this week.

Risk alerts are easy to read as lists of topics. The September 2015 alert is better read through its appendix. The topics were broad: governance, access, data loss, vendors, training and incidents. The requests were specific, and they asked for records that show controls running. [1]

Our argument is that the alert changed the unit of examination from the policy to the evidence of its operation. The appendix asked for policies, but it also asked for the date each former employee’s access ended, the systems without multifactor authentication and the incidents with their discovery dates. [1] In our reading, a firm could have every policy on the list and still be unable to answer the appendix.

Dates are the simplest form of that evidence. A pair of dates shows how long a control took. A list of exceptions shows how often it was bypassed. Neither can be produced by rewriting a policy. [1] In our reading, both are cheaper to keep from the start than to reconstruct later.

The approach also suits a security leader’s own needs. The same records that answer an examiner tell the CISO where controls are slow or bypassed. In our reading, the appendix described a management information set as much as an examination file.

That change had a direct consequence for technology leaders. Evidence of operation comes from systems: identity platforms, ticketing tools, logs, vendor files and incident trackers. If those systems did not keep the right records, no policy could supply them after the fact. [1]

We take the appendix in four parts, each a different kind of record:

  1. The access record: what the appendix asked about joiners, movers and leavers, and about authentication.
  2. The data record: what it asked about where customer information lives and how it leaves.
  3. The vendor record: what it asked about third parties with access to firm systems or data.
  4. The incident record: what it asked about plans, tests, alerts, incidents and losses.

Governance and training run across all four, so we treat them inside each part. The parts do not overlap: people, data, third parties and events. After them we look at who the alert reached and who it did not, set out the strongest objections and end with checks a CISO or chief compliance officer could run.

The access record: dates that prove a policy ran

Joiners, movers and leavers

The appendix asked for firm policies on establishing employee access rights, including role or group membership, and on updating or terminating access based on personnel or system changes. [1] That part could be answered from a policy manual.

It then asked for information demonstrating implementation. It named documentation evidencing the tracking of employee access rights, changes to them and any manager approvals for those changes. [1] It asked for former employees’ last date of employment and the date their access to firm systems was terminated. [1]

It asked the same for current employees reassigned to a new group or function: the date of reassignment and the date their access was modified. [1] Those pairs of dates are a test. A firm whose termination dates come from HR and whose access-removal dates come from IT can compare them. A firm that keeps only one of the two cannot.

In our reading, the request for paired dates was the most consequential line in the alert for identity teams. It turned the leaver process into something an examiner could measure. A policy that says access is removed promptly is a claim. A list of dates is evidence, and gaps in it are findings waiting to be written. [1]

Multifactor authentication as an inventory

The appendix asked for information on the systems or applications for which the firm uses multifactor authentication for employee and customer access. [1] It also asked for information on systems or applications for which the firm does not use it. [1]

That second request matters. It asked firms to list the exceptions as well as describe the rule. In our reading, an inventory of systems without multifactor authentication is a risk register in its own right. It shows where the firm has decided, or failed to decide, to accept a weaker control. [1]

Producing that list requires knowing every system with a login. Firms that had grown by adding hosted tools, vendor portals and legacy applications would find the inventory harder than the policy. The alert gave them a reason to build it. [1]

Failed logins, dormant accounts and exceptions

The appendix asked for policies on log-in attempts, failures, lockouts, unlocks and resets for perimeter-facing systems. [1] It asked how the firm enforces them and reviews those systems for failed log-in attempts, deactivation of access, dormant user accounts and unauthorized log-in attempts. [1]

It also asked for information on instances in which users, including employees, customers and vendors, received entitlements or access contrary to firm policy or without required authorization, and on any remediation. [1] That is a request for the firm’s own exception history.

A firm that never recorded such instances could not answer. A firm that recorded them would be showing an examiner its own lapses. In our reading, the second position is stronger. It shows the control detects problems, and the remediation record shows what happened next.

Remote devices and customer requests

The appendix asked about devices used to access firm systems externally, including firm-issued and personal devices. It asked about encryption and the firm’s ability to remotely monitor, track and deactivate those devices. [1]

It asked for customer complaints related to customer access and their resolution, and for policies on verifying the authenticity of customer requests to transfer funds. [1] Those requests joined the security function to the operations desk. A fraudulent transfer request is an access event and a payments event at once.

In our reading, the alert treated customer-facing authentication and funds-transfer verification as part of the same access record as employee logins. Firms that ran them in separate teams with separate evidence would need to bring the records together for an examination. [1]

Administrator accounts

The appendix’s examples of access policy included administrative management of systems. [1] It did not single out privileged accounts by name, but its requests on approvals, exceptions and leavers apply with the most force to them.

The R.T. Jones order shows why. The SEC found the adviser had limited access to the personal information on its server to two individuals with administrator status. [3] That limit did not stop an outside intruder from gaining access and copy rights to the data. [3]

In our reading, the lesson for access records is that the count of administrators is one data point among several. The others are how those accounts authenticate, where they can log in from and whether their activity is logged somewhere they cannot alter. [1] [3]

The data record: knowing where customer information lives

Data mapping and ownership

Under data loss prevention, the appendix asked for policies on enterprise data loss prevention and information on data mapping, with particular emphasis on understanding information ownership and how the firm documents personally identifiable information. [1]

It also asked about the systems, utilities and tools used to prevent, detect and monitor data loss relating to personally identifiable information and access to customer accounts. [1] The order of the requests is telling. The map comes before the tools.

A data map with owners is the precondition for every other data control. Monitoring cannot be tuned to data the firm cannot locate, and an owner cannot approve access to data no one is assigned to. In our reading, the alert asked firms to show that foundation, not to assume it. [1]

Classification with stated reasons

The appendix asked for policies on data classification. It asked for the types of classification, the risk level associated with each, the factors considered when classifying data and how those factors are weighed in classification decisions. [1]

That request asked for reasons as well as labels. A scheme that sorts data into low, medium and high without saying why cannot show an examiner that the most sensitive data received the strongest protection. The appendix used low, medium and high as its own example of risk levels. [1]

For a technology leader, classification with stated factors is also what makes automated controls defensible. A rule that blocks or allows a transfer is only as sound as the classification that triggers it.

Exfiltration channels

The appendix asked for policies on monitoring exfiltration and unauthorized distribution of sensitive information through channels such as email, physical media, hard copy or web-based file transfer programs, and for documentation of that monitoring. [1]

The main text of the alert framed the same point in volume terms. It said examiners may assess how firms monitor the volume of content transferred outside the firm by employees or through third parties, such as email attachments or uploads. [1]

In our reading, the request for documentation of monitoring is the key phrase. A firm that deployed a tool but kept no record of what it flagged and what was done could show the tool existed. It could not show the control operated. [1]

Patching as a data control

The alert’s main text placed patch management under data loss prevention, while the appendix listed it under governance and risk assessment. The main text said some breaches may have resulted from weaknesses in controls over patch management and system configuration. [1] The appendix asked for patch management practices, including prompt installation of critical patches and the documentation evidencing such actions. [1]

Patch evidence is a familiar audit artifact. What the alert added was the link to customer data. A missing patch on a system holding personally identifiable information was framed as a data loss risk as well as an IT hygiene issue. [1]

The appendix also asked for information on vulnerability scans, penetration testing policies and any related findings and remediation. [1] Findings without remediation records would show an examiner that the firm knew of a weakness. In our reading, that is a harder position than not having scanned, because it proves notice.

Data a firm did not need

The data map has a second use: finding data the firm should not hold. The R.T. Jones order is an example. The SEC found the adviser used the personal information of eligible plan participants to verify people enrolling through its website. [3]

To make that work, plan sponsors gave the adviser information on all their participants. The order found that although the firm had fewer than 8,000 plan participant clients, its web server held the personal information of over 100,000 individuals. [3]

The order also found the data was stored without modification or encryption on a third-party-hosted web server. [3] In our reading, a data map with owners and classifications would have shown a large store of non-client data on an internet-facing system. That is the kind of finding the appendix’s requests were built to surface. [1]

The point reaches beyond advisers. Any firm that accepts data from partners to verify customers can end up holding far more than its customer base. A map that counts records by the population they describe, as well as by system, shows that before an intruder does.

The vendor record: third parties inside the perimeter

Due diligence through to oversight

The appendix asked for vendor policies covering due diligence in selection, contracts and their approval, supervision, monitoring, tracking and access control, and any risk assessments, risk management and performance reports required of vendors. [1]

It asked for information on vendors with access to the firm’s network or data, including the services provided and the contractual terms on that access. [1] It also asked about vendors that provide cybersecurity services such as access controls, data loss prevention and management of personally identifiable information. [1]

In our reading, that second category is easy to overlook. A firm’s security vendors can hold some of its most privileged access. The appendix treated them as a distinct population to be inventoried and overseen. [1]

Change notices and contingency plans

Two requests in the vendor section look ahead rather than back. The appendix asked for sample documents or notices required of vendors before significant changes to their systems, components or services that could affect the firm’s data containing personally identifiable information. [1]

It also asked for written contingency plans with vendors covering conflicts of interest, bankruptcy or other issues that might put the vendor out of business or in financial difficulty. [1]

Both requests treat the vendor as a moving part. A contract that does not require notice of material changes leaves the firm to discover them. A relationship without a contingency plan leaves the firm exposed if the vendor fails. In our reading, the alert asked firms to write both into the relationship before an examiner asked. [1]

Training that reaches vendors

The training requests extended to third parties. The appendix asked for information on training provided to vendors or business partners on information security. [1] The main text said employees and vendors could be a firm’s first line of defense with proper training. [1]

It also said examiners may review how procedures for responding to cyber incidents under an incident response plan are integrated into regular personnel and vendor training. [1] That ties the vendor record to the incident record. A vendor that does not know the firm’s escalation path cannot use it.

For a firm that relies on outsourced help desks, hosting or administration, this is practical. The person who sees an early sign of compromise may work for a vendor. The alert asked whether that person had been trained in the firm’s procedures. [1]

Training records also need dates and names. The appendix asked for training dates, topics, methods and the groups of employees who took part. [1] A firm that tracks completion by person can show who missed training and what it did about it, which the 2017 observations later named as a gap at some firms. [4]

Vendor risk inside the firm’s risk assessment

The main text said examiners may assess how vendor relationships are considered in the firm’s ongoing risk assessment, and how the firm determines the appropriate level of due diligence for a vendor. [1]

That links vendor management back to governance. A risk assessment that lists internal systems but not the vendors that run or access them is incomplete. In our reading, the alert treated vendor access as part of the firm’s attack surface, to be assessed with the rest. [1]

Hosting providers as vendors

A hosting provider is a vendor in the appendix’s terms when it has access to the firm’s network or data. [1] The R.T. Jones server was third-party hosted, and the attack reached data the adviser stored there. [3]

Hosting arrangements split responsibilities. The provider runs the infrastructure; the firm decides what data goes on it and how it is protected. The order’s findings about missing encryption and a missing firewall concern the firm’s policies for its server and the data on it. [3]

In our reading, the appendix’s request for contractual terms on vendor access is where that split should be written down. Who patches, who monitors, who keeps logs and who tells whom about an intrusion are contract questions before they are technical ones. [1]

The incident record: plans, tests and the events themselves

Plans and tests

The appendix asked for policies and business continuity plans that address mitigating the effects of a cybersecurity incident and recovering from it. [1] It included policies on responsibility for losses associated with attacks or intrusions affecting clients. [1]

It asked about the firm’s process for testing or exercising its incident response plan, including frequency, reports and any remediation. [1] A plan that had never been exercised would produce an empty answer to that request.

In our reading, the request for test reports is the incident equivalent of the paired access dates. It asks the firm to show that the plan was used, and a written plan alone cannot show that. A tabletop exercise with a written report and tracked actions answers it. A plan on a shared drive does not. [1]

Alerts and incidents with dates

The appendix asked for system-generated alerts related to data loss of sensitive information or confidential customer records, with related findings and remediation. [1] It asked for incidents of unauthorized internal or external distribution of personally identifiable information, including the date of each incident, the discovery process, escalation and remediation. [1]

It asked the same for successful unauthorized access incidents: date, discovery, escalation and response. [1] These requests assume an incident register with fields for each stage. A firm that handled incidents by email thread would need to reconstruct those fields from memory.

The dates matter for the same reason as in the access record. The gap between incident and discovery, and between discovery and escalation, shows whether detection and governance worked. An examiner who has both dates can see the gap without the firm having to characterize it. [1]

Logs that survive the intruder

The incident record depends on logs, and logs can be attacked. The R.T. Jones order found that the cybersecurity firms the adviser hired could not determine the full nature or extent of the breach. [3] The reason the order gave was that the intruder had destroyed the log files covering the period of the intruder’s activity. [3]

The order found the firms could not, in the end, determine whether the stored personal information had been accessed or compromised. [3] The adviser then notified everyone whose information may have been compromised and offered identity monitoring. [3]

In our reading, that sequence shows what lost logs cost. Without them, the firm could not narrow the population at risk, so its notice and remediation had to cover everyone. Among its remedial steps, the order recorded a new firewall and logging system. [3] Keeping logs in a place an intruder on the server cannot reach is the control that would have preserved the answer.

Losses, reimbursement and insurance

The final requests concerned money. The appendix asked for the amount of actual customer losses associated with cyber incidents and the amount the firm reimbursed. [1] It asked whether the firm had cybersecurity insurance, what incidents it covered, whether claims had been filed and what had been recovered. [1]

Those requests reach finance and risk functions that may not sit with the CISO. In our reading, they also connect incident response to customer treatment. A firm that reimburses losses needs records tying each payment to an incident, and a firm with insurance needs evidence that meets the policy’s terms. [1]

The money questions also test honesty in the record. A firm that reports no incidents but has reimbursed customers for unauthorized transfers has a gap between its incident register and its ledger. In our reading, the appendix let an examiner find that gap by asking for both.

Governance across the four records

The governance section of the appendix ran across all four records. It asked for board minutes and briefing materials on cyber risks, incident response planning, actual incidents and vendor matters, where applicable. [1] It asked about the firm’s Chief Information Security Officer or equivalent and where cybersecurity positions fit in the organization. [1]

It also asked for periodic risk assessments and the remediation that followed them. [1] In our reading, the governance requests tested whether the four operational records reached the people accountable for them. An access gap that never appears in board materials, or an incident never briefed, tells an examiner as much as the gap itself.

An owner for each record

The four records can each have a different owner. HR owns departure dates, IT owns access removal, a data or privacy office owns the map, procurement owns vendor contracts and a security team owns incidents. Finance holds reimbursements and insurance.

The appendix asked for all of them at once. [1] In our reading, a firm preparing for that request needs one person accountable for assembling the set and for knowing where each piece lives. That role need not be senior. It needs to be named, and the owners of each record need to know it exists.

Naming that role also exposes gaps early. If no one can say who keeps the list of systems without multifactor authentication, the firm has learned something before an examiner does. [1]

Who the alert reached, and who it did not

Broker-dealers and investment advisers

The alert described examinations of registered broker-dealers and investment advisers. [1] Its footnote on Regulation S-P named the rule that required those firms to have written policies and procedures to protect customer information. [1]

For a large broker-dealer with dedicated security staff, the appendix was a checklist against existing evidence. For a small adviser with no security staff, it was a description of records the firm may never have kept. The alert said adequacy depended on each firm’s profile. [1]

In our reading, the R.T. Jones order, announced on September 22, 2015, showed the smaller end of that population. The SEC found that the adviser had about 8,400 client accounts and about $480 million in regulatory assets under management. [3] The rule applied to it as it did to larger firms. [3]

Banks and credit unions

The alert did not apply to banks or credit unions as such. Regulation S-P covers the securities firms it names, such as broker-dealers, investment companies and registered investment advisers. [1] The 2024 release describes separate incident response guidance that the banking agencies and the National Credit Union Administration issued in 2005. [6] A bank holding company with a broker-dealer subsidiary would see the alert apply to that subsidiary.

The SEC itself acknowledged the overlap later. Its 2024 Regulation S-P release discussed the banking agencies’ 2005 incident response guidance and said it had modified its amendments to promote consistency with other federal safeguard standards where that did not affect investor protection. [6]

In our reading, the 2015 appendix remains a usable evidence checklist for banks and credit unions, because its requests concern records any regulated firm would need. That is our suggestion, not a statement of what bank examiners require.

Vendors serving the industry

The alert did not reach vendors directly. It reached them through their clients. Requests for change notices, contingency plans, training and contractual access terms all land on the vendor when the firm tries to answer them. [1]

For a technology provider serving advisers or broker-dealers, the appendix was effectively a description of what clients would ask for. In our reading, vendors that could supply access logs, change notices and incident reports in a usable form made their clients’ examinations easier. [1]

Funds and, later, transfer agents

The 2015 document named broker-dealers and investment advisers as the examined population, while its footnote on Regulation S-P also listed investment companies among the firms the rule covers. [1] The 2017 observations reported on funds as well as broker-dealers and advisers. [4]

The population widened again later. The 2024 amendments extended the safeguarding requirements to transfer agents, according to the final rule’s summary. [6] In our reading, the appendix’s evidence requests therefore describe records that a wider set of securities firms now need.

Objections to this reading

“The alert was guidance, not a rule”

The alert said the Commission had expressed no view, that it was not legal advice and that it did not alter existing requirements. [1] A firm could reasonably say it created no new obligation.

That is accurate. The obligation came from Regulation S-P, which the alert cited, and the alert described how examiners would test it. [1] In our reading, a description of how a rule will be tested changes behavior even without changing the rule. The firms that met the appendix were the ones that had kept the records.

“Examiners would accept reasonable explanations”

A second objection is that examiners would accept a firm’s explanation where records were thin. The alert itself said adequacy depended on each firm’s facts and circumstances. [1]

That is true in part. But the appendix asked for specific records, and the 2017 observations later reported that a majority of firms’ information protection policies appeared to have issues, including policies that did not reflect actual practices. [4] In our reading, an explanation that is contradicted by the firm’s own records is weaker than no explanation.

“The six areas were already standard”

A third objection is that the six areas were familiar. The alert said its appendix tracked the NIST Framework released in February 2014, and OCIE had run a first round of cybersecurity examinations in 2014. [1]

The topics were familiar. The level of evidence was the change. The alert said this round would involve more testing to assess implementation. [1] The 2017 observations described the round in the same terms, as involving more validation and testing than before. [4]

“Records invite findings”

A fourth objection is uncomfortable but real. Keeping exception logs, incident registers and test reports creates documents that show weaknesses. A firm might prefer not to create them.

The alert closed that door. It asked for those records by name, including instances of access granted contrary to policy and incidents with their discovery dates. [1] A firm without them would be showing the examiner a different weakness: the absence of the record. In our reading, the R.T. Jones order showed the cost of that position, because the missing logs left the scope of the breach undetermined. [3]

“Small advisers cannot produce this”

A fifth objection is scale. A small adviser may have no identity platform, no ticketing system and no incident register. The appendix can look written for a firm with a security department. [1]

The document anticipated that. It said some factors may not apply to a particular firm’s business and that adequacy depends on each firm’s profile. [1] The records it asked for can be kept simply: a spreadsheet of leavers with two dates, a list of systems and their login methods, a log of incidents.

The R.T. Jones order also answers the objection. The rule applied to an adviser of that size, and the order found it had failed to adopt any written policies and procedures reasonably designed to safeguard its clients’ information. [3] In our reading, the gap at small firms was less about tools than about writing anything down.

“Dates can be gamed”

A sixth objection is that dates are easy to produce and easy to fake. A firm under examination could backfill a leaver list or tidy an incident register before handing it over.

The appendix guarded against that in part by asking for related records from different sources. Access-removal dates can be compared with system logs, incident dates with alerts, vendor notices with change records. [1] Records that come from separate systems are harder to align after the fact than a single document.

In our reading, the stronger defense is for the firm to keep those records in systems that stamp their own dates, such as ticketing and identity tools. A spreadsheet is acceptable evidence for a small firm. For a larger one, a record its own system generated carries more weight than one a person typed.

Checks a CISO or chief compliance officer could run

These checks are our suggestions, drawn from the 2015 appendix. They are not a complete statement of what Regulation S-P or any examiner requires. Firms should confirm their obligations with counsel. [1] [7]

  • Leaver dates: for each departure in the last review period, the HR end date and the access-removal date sit side by side, with gaps explained.
  • Mover dates: reassignments carry the date of the move and the date access changed.
  • MFA exceptions: there is a current list of systems without multifactor authentication, each with an owner and a decision.
  • Entitlement exceptions: access granted contrary to policy is logged with its remediation.
  • Data map: systems holding customer information are listed with owners and classifications that state their reasons.
  • Vendor file: each vendor with network or data access has contract terms on access, change notice and contingency.
  • Incident register: each incident records occurrence, discovery and escalation dates and remediation.
  • Exercise reports: incident response tests produce written reports with tracked actions.
  • Data minimization: any store of personal data about non-customers has an owner, a reason and a deletion date.
  • Hosting terms: contracts with hosting providers say who patches, monitors, keeps logs and reports intrusions.
  • Log retention: logs from systems holding customer information are kept outside those systems, so an intruder on the server cannot erase them.

The 2015 appendix asked firms to prove their controls with dates and records. [1] The checks above test whether those records exist before anyone asks. In our reading, that is the difference between an examination that confirms a program and one that discovers it.

With hindsight

On September 22, 2015, the SEC announced that R.T. Jones Capital Equities Management had agreed to settle charges that it failed to adopt required cybersecurity policies and procedures before a breach. [2] The SEC’s order found the firm violated Rule 30(a) of Regulation S-P, the safeguards rule. [3]

The order found that from at least September 2009 through July 2013 the firm stored clients’ and others’ personally identifiable information on a third-party-hosted web server without written policies and procedures to protect it. [3] It found that the policies did not include periodic risk assessments, a firewall for the server, encryption of the stored data or incident response procedures. [3]

It found that an intruder gained access and copy rights to the server in July 2013, and that the personally identifiable information of more than 100,000 individuals was rendered vulnerable to theft. [3] It said the consulting firms the adviser retained could not determine the full extent of the breach because the intruder had destroyed the log files. [3]

Without admitting or denying the findings, the firm agreed to a cease-and-desist order, a censure and a $75,000 penalty. [2] The order recorded its remedial steps, including appointing an information security manager, adopting a written policy, encrypting stored data and installing a new firewall and logging system. [3]

On August 7, 2017, OCIE published its observations from the 2015 initiative. It said staff had examined 75 firms, and that the exams involved more validation and testing of procedures and controls than before. [4] It said the exams were conducted between September 2015 and June 2016. [4]

The observations reported improvement and gaps. They said all broker-dealers, all funds and nearly all advisers examined had written cybersecurity policies on protecting customer records. [4] They also reported policies that were not reasonably tailored, or that did not reflect firms’ actual practices, such as required training that firms did not ensure was completed. [4]

In May 2024 the SEC adopted amendments to Regulation S-P. Its release said they require covered institutions to adopt incident response programs, including procedures to notify affected individuals. [5] The rule took effect on August 2, 2024, with compliance periods the release set at 18 months for larger entities and 24 months for smaller ones. [6]

The current text of the safeguards rule was checked on 2026-11-16. [7] It requires notice to affected individuals “as soon as practicable, but not later than 30 days,” after becoming aware that unauthorized access to or use of customer information has occurred or is reasonably likely to have occurred, subject to stated exceptions. [7] It also requires written policies and procedures reasonably designed to require oversight of service providers, including through due diligence and monitoring. [7] Those policies must be reasonably designed to ensure that a provider notifies the institution as soon as possible, but no later than 72 hours after becoming aware of a breach resulting in unauthorized access to a customer information system it maintains. [7]

Not every later proposal survived. On June 17, 2025, the SEC published a notice withdrawing certain proposed rules issued between March 2022 and November 2023, and said it did not intend to issue final rules on them. [8] The list included the proposed cybersecurity risk management rules for investment advisers and funds and for broker-dealers and other market entities. [8]

What held up from 2015 is the idea that examiners test evidence of operation. The 2017 observations confirmed the testing, and the 2024 amendments wrote incident response and recordkeeping into the rule itself. [4] [6] What did not hold up is the broader rulebook the SEC proposed for adviser and broker-dealer cyber risk management, which it withdrew. [8]

In our view, the R.T. Jones order is the best gloss on the alert. An adviser without written safeguarding policies, whose logs had been destroyed, could not say whether the stored information had been accessed. The appendix, published on September 15, had asked for exactly the records that would have answered. [1] [3]

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

  • SEC
  • OCIE
  • Regulation S-P
  • access management
  • incident response
  • broker-dealers