Morgan Stanley’s insider leak was an entitlement failure that exit controls could not catch
The January 2015 disclosure read as a data-loss story. Later findings showed four stages where one adviser’s downloads could have been stopped, and the firm’s best-documented controls guarded only the last two.
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
The January 2015 disclosure read as a data-loss story. Later findings showed four stages where one adviser’s downloads could have been stopped, and the firm’s best-documented controls guarded only the last two.
- The week On Monday, January 5, 2015, Morgan Stanley said it had begun telling some of its wealth management clients that an employee had stolen part of their account data.
- Entitlement: who could see which accounts The SEC’s account starts well before 2015.
- Aggregation: reports that bundle thousands of accounts The FTC described the same gap from a different angle.
- Exit: the controls the FTC credited The FTC’s letter is unusual because it lists the controls the agency gave the firm credit for.
- Discovery: finding the data outside The last link is how the firm learned of the theft.
- Where this argument is weakest The strongest objection is that entitlement cannot stop an insider from taking data he is entitled to see.
- A sequence for fixing the chain The practical question for a chief information security officer or chief technology officer is order.
- With hindsight The FTC closed its investigation first.
The week
On Monday, January 5, 2015, Morgan Stanley said it had begun telling some of its wealth management clients that an employee had stolen part of their account data. The firm’s statement gave three facts that framed the coverage that followed. Partial account information for up to 10 percent of all wealth management clients had been taken. Account names and numbers for about 900 clients had been posted on the internet for a short time. The data did not include passwords or Social Security numbers (Morgan Stanley).
The firm said it had found no evidence of economic loss to any client. The employee had been fired. Law enforcement and regulators had been told, and the firm said it was working with the authorities on an investigation. It would contact every potentially affected client, through the clients themselves and their financial advisors, and it had set up a dedicated information line. It was also adding security measures, including fraud monitoring on the affected accounts (Morgan Stanley).
The press filled in a number the statement did not give. Reports that day, citing CNBC and Bloomberg, put the affected group at about 350,000 clients, which matched the firm’s “up to 10 percent” (GovInfoSecurity) (Fortune). Fortune gave the scale of the business for context: at the end of 2013 the wealth management unit held about $1.9 trillion in client assets and had more than 16,700 representatives worldwide (Fortune).
How the firm found out was reported second-hand. A source close to the investigation told CNBC that the posted data was spotted on December 27, 2014, during a routine scan of suspicious websites. The same source said the firm traced it to the employee within 24 hours (GovInfoSecurity). Reporting that week also said the employee appeared to have been trying to sell the data. That claim came from people familiar with the matter, not from the firm, and the employee, according to reports citing people familiar with his account, denied posting the data or intending to sell it.
The firm did not name the employee. Later court records did, but the person was a private individual and the name adds nothing here. What matters for this article is the role: a financial advisor in the wealth management business, with ordinary day-to-day access to the firm’s client systems. He was not an administrator, a developer or a contractor with privileged rights. He was a front-line employee.
That detail is why the disclosure drew so much attention from security teams at banks and broker-dealers. The prior year’s headline breaches had been outside attacks. JPMorgan Chase had disclosed an intrusion affecting contact information for tens of millions of households in 2014. Sony Pictures was still dealing with its destructive attack, which the U.S. government had attributed to North Korea in December. The Morgan Stanley case was different. Nobody broke in. Someone who was allowed in took more than his job needed.
Most first-week commentary drew one conclusion: firms needed better data loss prevention, the tools that watch for sensitive data leaving the network. That was a reasonable first reading. The firm’s own statement pointed to a person, not a hack, and data loss prevention was the product category built for that case.
Two documents published much later changed the picture. In August 2015 the Federal Trade Commission closed its review and described the controls Morgan Stanley did have (FTC). In June 2016 the Securities and Exchange Commission settled a case describing the controls it did not have (SEC). Read together, they show that the January 5 statement was the visible end of a problem that started years earlier, in how the firm decided which employee could see which accounts. The rest of this article reads the week through those later findings.
The January 2015 disclosure is usually filed under insider threat, and the response most firms reached for was a stronger data loss prevention tool. We think that reading put the money in the wrong place. The record that emerged over the next 18 months shows a chain with four links, and the firm’s best-documented controls guarded only the last two.
The four links follow the path the data took. Each is a separate place where a firm can stop an insider, with its own owner, its own systems and its own evidence. Taken in order, they cover the whole route from a client record to a website outside the firm:
- Entitlement: which accounts an employee is allowed to see at all, decided by the authorization logic behind internal applications.
- Aggregation: the reports, queries and exports that bundle many accounts into one file, where screen-level checks often do not apply.
- Exit: the channels that carry data off the firm’s network, which data loss prevention, web filtering and device controls watch.
- Discovery: how the firm learns that data has left, from internal analytics or from finding it outside.
After the four links we take the strongest objections to this reading, and then set out the order in which a technology or security leader could fix the chain. The argument is narrow. Exit and discovery controls matter, but they are the most expensive place to stop an insider who was over-entitled in the first place, and the Morgan Stanley record shows why.
The facts we rely on come from three public documents. The firm’s January 5 statement gives the scale and the immediate response (Morgan Stanley). The FTC’s closing letter of August 10, 2015 describes the firm’s insider controls and the gap in them (FTC). The SEC’s June 2016 settlement describes the access failures over a longer period (SEC). Where the documents differ in emphasis, we say so.
Entitlement: who could see which accounts
The SEC’s account starts well before 2015. It found that from 2011 to 2014 the adviser accessed and moved data on about 730,000 accounts to a personal server at home. He did it through two internal web applications, or portals, that let employees view confidential customer account information (SEC).
The core finding concerned those portals. The SEC said the firm did not have effective authorization modules for more than ten years to restrict employees’ access to customer data based on a legitimate business need. It also found that the firm did not audit or test those modules, and did not monitor or analyze employees’ access to and use of the portals (SEC). The firm settled without admitting or denying the findings.
An authorization module is the piece of code that answers one question each time a page loads or a query runs: whether this user may see this record. It is distinct from authentication, which answers who the user is. Most large firms authenticate well. Authorization is harder, because the right answer depends on business facts that change every week.
Why wealth management entitlements are hard
In a wirehouse, the natural rule is that an advisor sees the accounts in his or her book. That rule breaks down fast. Advisors work in teams that share some households and not others. Client service associates support several advisors. Branch managers supervise everyone in the branch. Accounts move when an advisor retires, leaves or joins a team. Complaints, compliance reviews and operations work all need access across books.
Each exception is legitimate on its own. Together they push designers toward broad roles, because a broad role never generates a support ticket. An advisor who cannot see a teammate’s client during a vacation calls the help desk. An advisor who can see the whole region never complains. Over years, the path of least resistance produces access that is wide and badly understood.
The SEC’s wording suggests this kind of drift. It did not say the firm had no access policy. It said the modules meant to enforce business-need restrictions were not effective, and that nobody tested them (SEC). A policy can say “business need only” while the code says “anyone in this role.” Without testing, the gap stays invisible until someone exploits it.
What an entitlement control looks like when it works
A working entitlement model has three parts. The first is a data source of record for who owns or services each account. In wealth management that is usually the book-of-business or rep-code system, not the identity directory. Access rules should read from it directly, so that a team split or departure changes access the same day.
The second is authorization logic that is tested like any other critical code. Each portal should have automated tests that log in as a sample advisor and try to open accounts outside the book. The test passes only if the request is refused. These negative tests should run on every release and on a schedule, because configuration changes break them as often as code changes do.
The third is periodic certification with real data behind it. Managers sign off that their staff’s access is right. That exercise is often a rubber stamp. It works better when the reviewer sees what each person actually looked at as well as what they could look at. An advisor who opened thousands of accounts outside the book in a quarter is easy to spot on such a report.
Movers deserve special attention. Joiners get access through onboarding and leavers lose it through offboarding, both of which most firms run as formal processes. People who change teams, branches or roles often keep their old access and gain new access on top. Over a long career in one firm, an advisor or associate can accumulate rights to many books. A mover review that removes old access by default, and restores it only on request, closes much of that drift.
The evidence an examiner would ask for follows from those three parts. It includes the access rules and their data source, the test results for each application, the certification records with exceptions and their resolution, and the logs showing who viewed what. The SEC’s finding that the firm did not audit or test its modules for years is, in effect, a finding that this evidence did not exist (SEC).
Which rules apply to whom
The legal hook in this case was Regulation S-P, the SEC rule that implements the privacy and safeguarding provisions of the Gramm-Leach-Bliley Act for the firms the SEC oversees. Its Rule 30(a), the Safeguards Rule, requires broker-dealers, registered investment advisers and investment companies to adopt written policies and procedures reasonably designed to protect customer records and information (SEC).
Other financial institutions answer to parallel rules from other regulators. Banks and their trust departments follow the federal banking agencies’ interagency guidelines on safeguarding customer information. Federally insured credit unions follow the National Credit Union Administration’s security program rule. Non-bank lenders and many fintechs fall under the FTC’s Safeguards Rule. The duty is similar in each: protect customer information against unauthorized access, including by insiders. The examiners, the evidence they expect and the penalties differ. Firms should check with counsel on which rule governs each legal entity.
For a broker-dealer that is also part of a bank holding company, several of these regimes can apply at once to the same client data. This is one more reason to build entitlement controls once, at the application layer, and map them to each rule, rather than building separate controls for each regulator.
Aggregation: reports that bundle thousands of accounts
The FTC described the same gap from a different angle. Its staff found that the firm had a policy allowing employees to access only the personal data for which they had a business need. Yet the employee was able to reach client data despite that control, because the access controls applicable to a narrow set of reports were improperly configured (FTC).
That phrase, a narrow set of reports, points at the second link. A portal that shows one household at a time can enforce entitlement on each screen. A report that returns a list of accounts matching some criteria is a different code path. It often runs on a separate reporting engine, against a copy of the data, under a service account with broad rights. The entitlement check that protects the screen may not exist in the report at all.
How report paths escape screen-level checks
The pattern is common across financial services, and it is not unique to wealth management. Business users ask for reports that cut across accounts: all clients with a certain product, all households above a threshold, all accounts opened in a quarter. Developers build them quickly in a reporting tool. The tool authenticates the user, but filtering by book is left to the report’s author.
Some reports are built for staff who legitimately need wide access, such as compliance or operations. Those reports are then shared, cloned or exposed in a menu that other roles can reach. Nobody intends the advisor to run the compliance report. Nobody tests that the advisor cannot.
The result is a bulk channel with weaker controls than the screens around it. A user who would need thousands of clicks to page through accounts can get the same data in one file. That changes both the risk and the detection problem. Ten thousand page views look unusual. One report run looks routine.
Controls that fit the aggregation layer
Report authorization should use the same entitlement source as the screens, enforced in the data layer rather than in each report’s design. Row-level security in the database or the reporting platform does this. A report author then cannot forget to filter, because the filter applies to every query from that user.
Volume limits add a second check. A report that returns more rows than an advisor’s book size is a signal on its own. Firms can cap rows per run for front-office roles, require a second approval for larger extracts, and log every export with its row count and the user. Those logs are cheap to keep and easy to alert on.
Purpose matters too. Some firms tag each report with its intended audience and business purpose, and review the catalogue periodically. Reports nobody owns are retired. Reports with sensitive fields are reviewed before release. That governance work is dull, but it shrinks the attack surface more than any exit tool can.
Copies of the data outside the application
Reports are one aggregation path. Copies are another. Most firms copy client data from the core systems into a warehouse, a data lake or a set of departmental databases for analytics, marketing and supervision. Each copy tends to carry the data without the entitlement rules of the source application. Access to the copy is granted by role, and the role is usually broad.
Analysts, quantitative teams and vendors working on models often get the widest access of all, because they need history across the whole client base. That access may be justified. It should still be logged at the query level, limited to the fields the work requires and reviewed as often as front-office access. Masking or tokenizing account numbers in analytic copies removes much of the value of a stolen extract without hurting most analysis.
The simplest test is to list every place a full client list could be pulled from in one query. In many firms the answer is longer than the security team expects. Each entry on that list is an aggregation point that needs its own control and its own owner.
What it costs
The trade-off is real. Front-office staff use reports to prospect, service and plan. Tighter row limits mean more requests for exceptions and slower answers for clients. Row-level security on a legacy reporting platform can be a sizeable project. A firm should expect pushback from the business and plan for an exceptions process with a clear owner and a turnaround target.
The cost of not doing it is also concrete. Under Regulation S-P, the question an examiner asks is whether the firm’s policies were reasonably designed to protect customer information. A report path that ignores the firm’s own business-need rule is hard to defend as reasonable once someone has used it.
Exit: the controls the FTC credited
The FTC’s letter is unusual because it lists the controls the agency gave the firm credit for. Staff noted a policy limiting employees to data they needed. They noted that the firm monitored the size and frequency of data transfers by employees. It also prohibited the use of USB and other devices to take data out, and blocked employee access to certain high-risk web applications and sites (FTC).
Those are the standard exit controls of the period. Volume monitoring looks for unusually large or frequent transfers. Device control stops copying to removable media. Web filtering blocks categories of sites, such as personal file-sharing and webmail, where data could be uploaded. The FTC said these comprehensive policies were one reason it closed the investigation (FTC).
How the data still left
The FTC’s letter describes the route. The employee transferred data from the firm’s network to a personal website accessed at work, and from there onto personal devices (FTC). The letter adds a pointed observation: as employees increasingly use personal websites and online applications, companies should deploy appropriate controls for the risks of broad access to such resources on work devices.
The mechanism is easy to reconstruct in general terms. Web filtering works by category. A vendor classifies sites as file sharing, webmail, social media and so on, and the firm blocks the risky categories. A small personal website that no vendor has seen has no category. Many firms allowed uncategorized traffic by default in 2015, because blocking it caused help-desk calls whenever staff visited a legitimate but obscure site.
Volume monitoring has a parallel blind spot. Thresholds are set to catch a large one-time transfer. An insider who moves data in small pieces over months stays under them. The SEC’s period, 2011 to 2014, suggests the data left over a long time rather than in one event (SEC). A rule tuned for a smash-and-grab will not fire on a slow drip.
What exit controls can and cannot do
Exit controls are worth having. They raise the cost of casual theft, they produce evidence after the fact, and regulators expect them. But they face a structural problem that entitlement controls do not. They have to inspect every channel out of the firm, and the number of channels keeps growing. Each new cloud application, messaging tool or personal device is a new path to cover.
Entitlement works in the opposite direction. It narrows what any one person can take, whatever channel they use. An advisor who can see only the accounts in his book can steal only those accounts. That would still be a loss, but a loss of hundreds of records rather than hundreds of thousands. The FTC’s list of controls is a good list. It was protecting a pool of data far larger than any one employee needed.
Personal devices and remote access
The FTC’s letter mentions personal devices as the final stop (FTC). In 2015 most advisors worked from branch desktops. Since then, remote work, mobile access and personal devices have multiplied the routes out. A control set designed for a branch network does not cover an advisor reading client data on a home laptop through a browser session.
Virtual desktop and secure browser approaches help by keeping the data on firm infrastructure and limiting copy, print and download. They also create their own gaps, such as screenshots and phone cameras, which no technical control fully closes. That residual gap is one more argument for reducing what each person can see in the first place.
Tightening the exit layer anyway
Several practical steps would have narrowed this specific path. Firms can block uncategorized and newly registered domains on devices that can reach client data, with a fast exception process for business sites. They can apply content inspection that recognizes account number formats as well as file size. They can route all web traffic from client-data systems through a proxy that logs uploads, and alert on any upload of structured data to a personal destination.
Each step has costs. Blocking uncategorized sites generates complaints. Content inspection produces false positives, and someone must triage them. Proxy logging raises employee privacy questions, which firms should address in policy and with counsel in each jurisdiction where staff work. Even together, these steps do not remove the need to fix entitlement, which is why we put exit third in the sequence, not first.
Discovery: finding the data outside
The last link is how the firm learned of the theft. By the account reported that week, the firm spotted the posted data on December 27, 2014, during a routine scan of suspicious websites, and identified the employee within 24 hours (GovInfoSecurity). That is fast work once the data was visible. It is also a sign of what had not worked earlier.
The SEC’s release adds a detail that matters. It says a likely third-party hack of the employee’s home server led to portions of the data being posted online, along with offers to sell more (SEC). If that account is right, the firm found out because someone else attacked the thief. Without that second event, the data could have stayed on a home server indefinitely.
Internal detection that did not fire
The SEC found that the firm did not monitor or analyze employees’ access to and use of the two portals (SEC). This finding concerns the internal detection layer. It would have compared each user’s access pattern with a baseline: how many accounts an advisor of that type usually opens, how many are outside the book, at what hours, through which reports.
Tools that do this are often sold as user and entity behavior analytics. The method does not need a special product. It needs application logs that record who viewed which account, a feed of each user’s assigned book, and a simple comparison. An advisor who views many accounts outside the book is an exception worth a look. An advisor who runs the same wide report every week is a stronger signal.
The obstacle is usually logging rather than analytics. Many internal portals in 2015 logged logins and errors but not record-level views. Without per-record logs, a firm cannot detect an insider, and cannot tell regulators afterward which clients were affected. That second point turns a security gap into a notification problem.
External discovery as a backstop
Scanning paste sites and underground markets for the firm’s data is a reasonable backstop. It found this case. It also has limits. It only finds data that someone chooses to post, and by then the data is already out. For client notification and regulatory reporting, it starts the clock late.
Scoping who was affected
Once a theft is known, the first hard question is scope. The firm must establish which clients’ data the person took. Without record-level logs, the honest answer is everyone the person could have seen. This is likely why the firm’s first statement spoke of up to 10 percent of wealth management clients rather than a precise list (Morgan Stanley). The number of people a firm must contact is set by the breadth of the entitlement, not the size of the theft.
Good logs change that. A firm that can show exactly which accounts a user viewed or exported can notify those clients and no others. It can also tell regulators with confidence what was taken. Over-notification costs money and trust. Under-notification creates legal exposure. Both come from the same missing data.
Scoping also needs forensic access to the insider’s devices and accounts. The employment agreement, the acceptable-use policy and the device management settings should make clear that the firm can preserve and examine work devices and firm accounts. Firms should check with counsel on what they may do with personal devices and personal accounts, which is far more limited.
A firm should treat external discovery as evidence that the earlier links failed, and run a post-incident review against each of them. The review should establish which entitlement let the person see the data, which report bundled it and which channel carried it out. The answers become the remediation plan, and examiners will expect to see it.
Response once the data is found
Morgan Stanley’s first public steps were standard and sensible. It fired the employee, told law enforcement and regulators, contacted clients through their advisors, set up a call line and added fraud monitoring (Morgan Stanley). The scale made those steps hard. Contacting a large share of a wealth management client base through advisors means thousands of personal conversations, each of them a test of the relationship.
Whether the theft triggered legal notice duties depended on the data. Account names and numbers without access codes fall outside some state definitions of personal information and inside others. In 2015 there was no federal rule that set a single standard for a broker-dealer’s customer notice. Firms facing a similar event should map their data elements against each state’s law, with counsel, before an incident rather than during one.
Where this argument is weakest
The strongest objection is that entitlement cannot stop an insider from taking data he is entitled to see. An advisor with a large book can still copy that book. The point is fair, and the residual risk is real. It is also a much smaller one. The difference between a book of a few hundred households and a firm-wide pool of hundreds of thousands is the difference between a serious incident and a regulatory case.
A second objection is that tight entitlement damages service. Clients call when their advisor is away. Teams cover for one another. Operations staff fix problems across books. Overly strict rules create workarounds, such as shared logins, that are worse than broad access. The answer is not to give up on entitlement but to design for coverage: time-bound delegation, logged break-glass access and fast exception handling. Those are well-understood patterns.
A third objection points to the FTC’s decision, and it deserves a fuller answer than the others. The FTC looked at the same events, credited the firm’s controls and closed its review (FTC). If one regulator found the controls reasonable, perhaps the problem was a narrow misconfiguration, quickly fixed. We read the record differently. The FTC described a narrow set of misconfigured reports. The SEC described authorization modules that were not effective for more than ten years and were never tested (SEC). The two findings are consistent with each other. The specific hole was narrow. The absence of testing that would have found it was not.
The two agencies were also asking different legal questions. The FTC was deciding whether to bring a case under Section 5 of the FTC Act, which turns on whether practices were unfair or deceptive (FTC). The SEC was applying a rule written for its registrants, which asks whether written policies and procedures were reasonably designed to protect customer information (SEC). A firm can have impressive controls and still fail the second test if the controls that matter were never verified. For a broker-dealer, the SEC’s test is the one that governs.
A fourth objection concerns employee privacy. Record-level logging, behavior analytics and web inspection all watch staff closely. Employees and, in some countries, works councils and data protection authorities will ask whether the monitoring is proportionate. That concern is legitimate, especially for firms with staff in Europe. The entitlement-first approach helps here too. Narrow access reduces how much monitoring is needed to manage the risk, because there is less to watch for.
A fifth objection is cost. Rebuilding authorization across legacy portals and reporting tools is expensive, and the benefit is a loss that may never occur. Exit tools, by contrast, are bought, installed and reported on in a quarter. That is a fair description of the budget dynamics. It is also why so many firms bought exit tools first. The later enforcement record, including the SEC’s penalty and its public findings, is the cost side of that choice.
A sequence for fixing the chain
The practical question for a chief information security officer or chief technology officer is order. Every link needs work, and no firm can fix all four at once. The sequence below assumes a mid-sized or large firm with legacy applications and a twelve-month horizon. Smaller firms can compress it. Our sequence follows the four links but starts where each dollar buys the most reduction in risk, which is entitlement and aggregation, and leaves exit and discovery for tuning rather than rebuilding.
First quarter: find the wide paths
Start with an inventory of every application and report that can show client data across more than one book. For each, record the authorization logic, its data source and whether it has negative tests. The first result is usually a short list of high-risk paths: a legacy portal with role-based access only, a reporting platform with no row-level security, a data warehouse with broad analyst access.
At the same time, turn on record-level logging where it is missing, even before fixing the access. Logs make the later work measurable, and they make any incident in the meantime easier to scope.
Second and third quarters: close the paths
Fix entitlement in the highest-risk applications first, reading from the book-of-business system. Add row-level security to the reporting layer. Cap rows per run for front-office roles. Retire reports with no owner. Build the negative tests into release pipelines and run them on a schedule.
Assign an owner for each fix. Entitlement logic is usually owned by the application team, but the rules belong to the business. A wealth management chief operating officer or head of supervision should sign the rules. The security team should verify them.
Fourth quarter and after: tune exit and discovery
With the pool of reachable data smaller, exit controls become more effective. Block uncategorized domains on client-data devices. Add content inspection for account numbers. Build the behavior analytics on top of the new logs. Rehearse the incident response plan against an insider scenario, including who decides on client notice and how fast.
What to report to the board
Boards respond to measures. Useful ones include the share of client-data applications with tested authorization and the number of users with access beyond their book. Others are the number of large exports per month and the time to remove access after a role change. Each measure maps to one link in the chain. Together they show whether the firm is reducing what any single insider could take.
Vendors, contractors and service providers
The same chain applies to people outside the firm who can see client data. Offshore operations teams, software vendors with support access and contractors on data projects all hold entitlements. Their access is often granted by project and rarely removed when the project ends. Contracts should require the provider to enforce least-privilege access, keep record-level logs and give the firm those logs on request.
The 2024 amendments to Regulation S-P added service-provider oversight to the rule for the firms it covers, alongside the incident response program and customer notice duty (SEC). A firm that has fixed its own entitlement chain but not its providers’ has fixed half the problem. Vendor reviews should ask the same four questions this article asks of the firm: who can see what, through which bulk paths, by which exits, and how the firm would know.
Different firms, different starting points
The sequence applies across firm types, but the starting point differs. A large wirehouse has many legacy portals and a big reporting estate, so aggregation is usually the biggest job. An independent registered investment adviser often runs on a custodian’s platform and a few software-as-a-service tools. Its entitlement controls live mostly in vendor settings, and its main task is to configure and review them. A bank trust department shares systems with the bank, so its client data is often reachable by bank staff with no trust role. A credit union with an investment program run through a third-party broker-dealer has to check that partner’s controls as well as its own.
With hindsight
The FTC closed its investigation first. Its staff letter of August 10, 2015 said the firm had comprehensive policies against insider theft and that the employee got through because access controls on a narrow set of reports were misconfigured. The letter added that the firm fixed the problem promptly once it came to the company’s attention (FTC). The letter also said, as such letters do, that closing the investigation was not a finding that no violation occurred.
The criminal case ended later that year. The employee pleaded guilty in federal court in Manhattan to a charge of unauthorized access to a computer. In December 2015 he was sentenced to 36 months’ probation and ordered to pay $600,000 in restitution, as the SEC later recounted (SEC).
The SEC’s case came in June 2016 and set the record that has shaped how this incident is read. The agency found that Morgan Stanley Smith Barney violated Rule 30(a) of Regulation S-P, the Safeguards Rule. It found that the firm’s two portals lacked effective authorization modules for more than ten years, that the modules were never audited or tested, and that employee access was not monitored. The firm agreed to pay a $1 million penalty without admitting or denying the findings. The former employee agreed to a separate industry bar (SEC).
Two points from that order held up well. First, the SEC treated a failure of access control, not a failure of exit control, as the violation. Second, it treated the absence of testing as part of the failure. That framing is the useful one for firms today. The question an examiner can ask is what the access policy says and how the firm proves its code enforces it.
A later Regulation S-P case against the same firm concerned a different stage in the data’s life. In September 2022 the SEC announced that Morgan Stanley Smith Barney would pay $35 million over failures to protect the personal information of about 15 million customers. The case concerned the disposal of hard drives and servers, starting as early as 2015, through a moving and storage company with no data destruction experience. Some devices were resold online with customer data on them. The SEC also cited 42 missing servers and encryption that had not been turned on (SEC). The incident was unrelated to the 2015 insider, but the theme was the same. Controls existed on paper and failed in operation where nobody checked.
The rule itself changed in 2024. On May 16, 2024 the SEC adopted amendments to Regulation S-P. They require broker-dealers, funding portals, investment companies, registered investment advisers and transfer agents to keep an incident response program designed to detect, respond to and recover from unauthorized access to customer information. They also require notice to affected individuals as soon as practicable, and no later than 30 days after the firm learns of an incident that compromised sensitive customer information (SEC). Larger firms had 18 months from publication to comply and smaller firms 24 months, which put the dates in December 2025 and June 2026. By the time of writing, both groups should be operating under the amended rule.
Those amendments close one gap the 2015 case exposed. In January 2015 a broker-dealer’s notice to customers was governed mostly by a patchwork of state laws, whose definitions did not always cover account numbers without access codes. A 30-day federal notice duty now applies to the SEC-registered firms the rule covers. It does not change what the firm must do to prevent the theft, but it makes the discovery link and the response plan an examination topic.
Some things did not hold up. The early coverage framed this as a data loss prevention problem, and the industry bought accordingly. Exit tooling has a place, but the later enforcement record against this firm kept pointing back to entitlement, testing and operations. Another early claim, that the employee intended to sell the data, was never established in the public record we could find, and we have not relied on it.
What remains is the shape of the chain. The four links in this article, entitlement, aggregation, exit and discovery, still describe how client data leaves a financial firm through an insider. The cheapest place to cut the chain is still the first link. Most firms still spend most of their insider-risk budget on the third.
Spotted an error? Tell us and we will check it and correct the article where needed.