EDGAR’s hacked test filings put the pre-release path inside the perimeter

On January 15, 2019, the SEC charged nine defendants with trading on earnings results hacked from EDGAR test filings. Every place a draft waits before publication, including the regulator’s own test channel, belonged inside issuers’ and broker-dealers’ information barriers.

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

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

On January 15, 2019, the SEC charged nine defendants with trading on earnings results hacked from EDGAR test filings. Every place a draft waits before publication, including the regulator’s own test channel, belonged inside issuers’ and broker-dealers’ information barriers.

  1. The week On January 15, 2019, the Securities and Exchange Commission announced charges against nine defendants for taking part in what its press release called a previously disclosed scheme to hack into the SEC’s EDGAR system and extract nonpublic information for illegal trading.
  2. The issuer’s choice: what goes into a test copy The SEC’s release made a point that is easy to pass over.
  3. The intermediary’s copy: newswires and agents The EDGAR scheme was, by the SEC’s account, the second phase of a longer effort.
  4. The regulator’s copy: EDGAR’s test channel and return copies The third holder on the path was the SEC itself.
  5. The clock: the window between extraction and release Every copy on the pre-release path is dangerous only for as long as the information is not public.
  6. The trading end: what broker-dealers could see Information stolen from a pre-release system is worth something only if it can be traded.
  7. A control set for the pre-release path The four parts of the path and the trading end translate into a short set of controls.
  8. Objections, and where the argument stops The strongest objection is that the EDGAR case was a failure of the SEC’s software, and that drawing lessons for issuers and broker-dealers blames the victims.
  9. With hindsight The October 2, 2017 update had said that the EDGAR system had been undergoing modernization efforts, and listed a focused review of EDGAR, with outside consultants and an increased focus on cybersecurity, as one of its work streams.

The week

On January 15, 2019, the Securities and Exchange Commission announced charges against nine defendants for taking part in what its press release called a previously disclosed scheme to hack into the SEC’s EDGAR system and extract nonpublic information for illegal trading. The release described the defendants as a Ukrainian hacker, six individual traders in California, Ukraine and Russia, and two entities. It also named four relief defendants, people the SEC said had profited when the defendants used their brokerage accounts to place trades. The same day, the release said, the U.S. Attorney’s Office for the District of New Jersey announced related criminal charges.

The target was a part of EDGAR the investing public never sees. The release explained that issuers can elect to submit “test files” ahead of their official filings, to make sure EDGAR will process the filings as intended, and that some issuers chose to put nonpublic information in those test filings, such as actual quarterly earnings results not yet released. The SEC’s complaint, filed that day in the U.S. District Court for the District of New Jersey as Civil Action No. 19-cv-505, described test filings as draft versions meant to ensure that a filing is in the correct format, free from errors and will be accepted, and said they are not meant for public dissemination. A filer could request a “return copy” to view the test filing it had submitted. During the period in the complaint, those return copies were kept on SEC servers in Middlesex County, New Jersey, and were not public.

According to the complaint, the hacker circumvented EDGAR controls that required user authentication, reached nonpublic EDGAR logs containing user and document identification information for test filings, and used that information to reach return copies. The complaint said he misrepresented himself as an authorized EDGAR filer, used aliases to hide his control of an IP address and a related domain, and that he or others working with him sent spoofed emails, made to look as if they came from SEC security personnel, carrying malware-infected documents that infected several SEC workstations. Beginning on or about May 3, 2016, and continuing until at least October 20, 2016, the complaint alleged, they downloaded thousands of then-unpublished return copies of test filings. On May 18, 2016, a server the complaint called the “Exfiltration Machine” began downloading return copies on an automated basis.

The release said the traders traded before at least 157 earnings releases from May to October 2016 and generated at least $4.1 million in illegal profits. Steven Peikin, co-director of the Enforcement Division, said the trader defendants were alleged to have used an offshore entity and nominee accounts to place trades, and that the staff’s analysis of the trading exposed “the common element behind their success.” The complaint also described trading in accounts held in other people’s names. His co-director, Stephanie Avakian, said such schemes were a constant risk for any organization that holds valuable information.

The conduct was not new territory for the SEC. The release said the hacker and some of the traders had been involved in a similar scheme to hack newswire services, and that the SEC had charged them for that in 2015. The SEC’s August 11, 2015 release in that case announced fraud charges against 32 defendants and alleged more than $100 million in illegal profits from earnings announcements stolen from newswires before release. The 2019 complaint said the newswire hacking ran from 2010 until 2015 and reached over 100,000 draft press releases.

The EDGAR intrusion itself was also public before the week. In a statement dated September 20, 2017, Chairman Jay Clayton said the Commission had learned in August 2017 that an incident detected in 2016 may have provided the basis for illicit gain through trading, and that a software vulnerability in the test filing component of EDGAR, “patched promptly after discovery,” had been exploited and resulted in access to nonpublic information. An SEC update dated October 2, 2017 added that one accessed test filing contained the names, dates of birth and Social Security numbers of two individuals, and set out five work streams for review and remediation, including a focused review of EDGAR.

Insider-trading cases are easy to picture as a story about a person: an executive who tips a friend, a banker who trades ahead of a deal. The January 2019 EDGAR case described something else. Nobody inside the issuers was alleged to have leaked anything. The information was taken from copies of earnings results that the issuers had placed, on purpose and for sound operational reasons, in someone else’s system before publication. The previous phase of the same scheme had done the same thing to newswire services.

Our argument is that the case moved the boundary of the information barrier. Information barrier policies are written around people: who may know material nonpublic information, and who may not. The EDGAR and newswire cases showed that what needs controlling is every copy of the information and every system that holds a copy, from the moment a draft leaves the drafting team until the moment the market has it. That pre-release path runs through systems the issuer does not own, including, in this case, the regulator’s own. A security program that stops at the issuer’s network edge does not cover it.

We break the pre-release path into four parts that do not overlap: the issuer’s own choice about what goes into a draft copy; the intermediaries that hold the release before the market does; the regulator’s channel, where EDGAR test filings and their return copies were kept; and the clock, meaning the length of time any copy waits before publication. A fifth section turns to the other end of the scheme, the brokerage accounts the trades ran through, because broker-dealers carry their own obligations there. The last section takes the strongest objections. Throughout, this is analysis, not legal advice. Issuers and broker-dealers should check specific obligations with counsel.

The issuer’s choice: what goes into a test copy

The SEC’s release made a point that is easy to pass over. Issuers “can elect” to submit test filings, and some “elected” to include nonpublic information in them, such as actual quarterly earnings results not yet released. The complaint used the same language: filers “may also elect” to submit test filings, and test filers “sometimes elected” to include material nonpublic information, giving drafts of forthcoming Forms 8-K as an example. Twice, the word is a choice.

A test filing exists to check format and acceptance. The complaint said so directly: draft versions meant to ensure that a filing is in the correct format, free from errors, and will be accepted for filing. None of those purposes needs the real numbers. A test of an earnings 8-K checks that the exhibit attaches, the tags validate and the submission type is right. Placeholder figures would test the same things. The choice to send real figures was a convenience, and the case put a price on it.

Of every control on the path, this one costs the issuer least, and it belongs to the issuer alone. An issuer cannot patch the regulator’s software or audit a newswire’s servers. It can decide that no copy of a release leaves its own environment with the real figures until the moment it has to. The same reasoning reaches beyond EDGAR. Any rehearsal of a publication step, whether a test upload, a formatting proof or a check of how a release will render, can run on dummy data.

Drafts carry more than earnings

The October 2, 2017 update showed a second kind of exposure in the same channel. The SEC said a test filing accessed in the intrusion contained the names, dates of birth and Social Security numbers of two individuals, and that staff were reaching out to offer them identity theft protection and monitoring. That filing was not an earnings release. It shows that the pre-release path carries personal data as well as market-moving data, and that a test copy can hold whatever the final document holds. A data-minimization rule for test submissions protects both.

Who decides, and where it is written down

For a public company, the decision about what goes into a test submission sits with whoever runs the filing process: financial reporting, the corporate secretary’s office, investor relations or an outside agent. For the control to hold, it needs an owner and a written rule. A disclosure committee charter or a filing procedure can state that test submissions use placeholder values for financial results and personal data, and that real figures enter only the live submission. The control is simple to audit: sample past test submissions and check what they contained.

Banks and broker-dealers that are themselves public companies face this choice twice. They file their own reports, and they hold drafts of their clients’ information in advisory, underwriting and research work. The second role is the one their information barriers were built for. The first can fall outside those barriers wherever the firm’s own earnings release is handled by finance and investor relations teams that the barrier policies were not written around. The EDGAR case is a reason to put the firm’s own pre-release path inside the same program.

What the issuer cannot choose

Not everything is in the issuer’s hands. A live filing has to carry the real numbers, and a press release has to reach the newswire before it can be distributed. The issuer can choose what goes into rehearsal copies, and it can choose when the real copy leaves. It cannot choose how the receiving system protects that copy. The next two sections deal with those receiving systems, and the one after with timing.

The practical test for this part of the path is a single question an internal auditor can ask of the filing team: in the last four quarters, did any copy of the earnings release or the earnings 8-K, with real figures, exist outside our systems before the publication time, and if so, why? The answer will be yes for the live filing and the newswire submission. It should be no for everything else.

The intermediary’s copy: newswires and agents

The EDGAR scheme was, by the SEC’s account, the second phase of a longer effort. The 2019 complaint said the hacker was involved in hacking at least three newswire services, one after another, and that from 2010 until 2015 he and others accessed over 100,000 draft press releases before they were published. The SEC’s 2015 release in the newswire case alleged that two hackers used proxy servers to mask their identities and posed as newswire service employees and customers, and that they sent stolen data to traders in several countries through a secret web-based location.

The complaint described why newswires were worth attacking in terms that apply to any intermediary. Issuers, it said, “routinely provide draft press releases” to the newswire services, and these contain material nonpublic information. Until publication, the information is nonpublic. So for each release there is a window between the moment the company provides the draft and the moment the newswire publishes it. The complaint put those windows at between a few minutes and a few days.

The 2015 release gave one example of how narrow a window could be and still be used. On May 1, 2013, it alleged, the hackers and traders moved in the 36-minute period between a newswire’s receipt and release of an announcement that a company was revising its earnings and revenue projections downward. Ten minutes after the company sent the confidential release to the newswire, the release said, traders began selling the stock short and selling contracts for difference, and realized $511,000 in profits when the price fell.

The intermediary is inside the perimeter whether the contract says so or not

From the issuer’s side, a newswire, a financial printer or a filing agent is a vendor. From the attacker’s side, it is a concentration point: one system holding the unreleased results of every client that uses it. The newswire case showed the attacker’s view was the accurate one. A breach of one intermediary exposed the drafts of every issuer using it, and none of those issuers had done anything wrong inside their own walls.

For an issuer, that means vendor oversight for these intermediaries should look like oversight of any third party holding the firm’s sensitive data. The questions are familiar from any third-party risk program: how the vendor authenticates the people who upload and view drafts, how long drafts stay on its systems, who on its staff can see them, how it logs access, and how quickly it will tell the issuer if a draft may have been exposed. The difference is the deadline. The 2015 and 2019 records show stolen releases traded within minutes, and the harm fell on the market as well as on the issuer.

Agents who submit on the issuer’s behalf

Where a third party submits EDGAR filings on an issuer’s behalf, that party holds the draft too, and holds the credentials that let it into EDGAR. The SEC’s 2024 EDGAR Next release, discussed in the hindsight section, refers to individuals acting on behalf of EDGAR filers. The 2024 rule was the step that tied filing access to individual Login.gov credentials. An issuer relying on an agent in 2019 had two things to secure at that agent: the copy of the draft and the access codes.

Contracts are the issuer’s main tool here. A filing or distribution agreement can require placeholder data in test submissions, set a maximum time between receipt of a final release and its scheduled publication, require prompt notice of any suspected access to unreleased material, and give the issuer the right to ask how access is controlled. Each clause is familiar from contracts with vendors holding sensitive data. What the EDGAR and newswire cases add is the reason to apply them to the vendors that handle publication, which a vendor program might otherwise rate as low risk because they handle information that is about to become public anyway.

About to be public is not the same as public

That last point carries the argument. A release that will be public at 4:05 p.m. is, at 3:40 p.m., information that the 2015 and 2019 cases show can be traded for profit. A vendor risk rating that asks only whether the data will eventually be published will score a newswire or filing agent too low. The rating should ask what the data is worth for the period it is not yet public, and for how long the vendor holds it in that state.

The regulator’s copy: EDGAR’s test channel and return copies

The third holder on the path was the SEC itself. The complaint set out a chain of steps. The hacker circumvented EDGAR controls that required user authentication, reached nonpublic EDGAR logs with user and document identification information for test filings, and used what he found there to reach return copies. He misrepresented himself as an authorized filer. He or others working with him sent spoofed phishing emails, made to look as if they came from SEC security personnel, with malware-infected attachments that infected several SEC workstations. He used aliases and IP addresses that appeared to be in different parts of Europe. On May 18, 2016, a server began downloading return copies on an automated basis.

Chairman Clayton’s September 20, 2017 statement described the root cause in one sentence: a software vulnerability in the test filing component of EDGAR, which was “patched promptly after discovery,” had been exploited and resulted in access to nonpublic information. The same statement said the Commission learned in August 2017 that the incident, detected in 2016, may have provided the basis for illicit gain through trading.

Detected as an intrusion, understood as a market event

For a security team, those two dates deserve close attention. The intrusion was detected in 2016 and the vulnerability patched. The possibility that it had provided a basis for illicit trading was learned in August 2017. The October 2, 2017 update said that one of five work streams was an internal review of the procedures followed in response to the intrusion, and that staff would work to enhance escalation protocols for cybersecurity incidents to give greater agency-wide visibility of potential vulnerabilities and attacks. The record we read does not explain the timing, and we do not speculate. What the record does show is that a security response that closes a vulnerability does not, by itself, answer the business question of what the attacker took and whether anyone used it.

For any organization that holds other people’s market-moving information, that is a design point for incident response. When a system holding pre-release information is breached, the response plan should include a step that asks which documents were reachable, whose information they contained, when each was published, and whether anything unusual happened in the market for those securities before publication. That step needs people who understand both the logs and the market, and a route to the people who can answer for the market side.

The return copy problem

The return copy feature is a useful case study in how a convenience creates a store of sensitive data. A filer could request a return copy to view the test filing after submitting it. The complaint said the return copies were kept on SEC servers. So the system held, for some period, copies of unreleased results that existed only because filers wanted to look at their test submissions. Every system that lets a user preview, render or proof a document before publication creates a similar store. Each is worth asking about: who can reach it, how long it is kept, and whether it needs the real data at all.

The complaint also described the logs the hacker reached as containing user and document identification information. In this case, the logs were the map that led to the documents. A classification scheme for a pre-release system should cover the documents and also the metadata that points to them.

What issuers could and could not control at the SEC

No issuer could audit EDGAR. The point of including the regulator in this analysis is not that issuers should have secured the SEC’s systems. It is that the issuer’s threat model for its own unreleased results should have listed the regulator’s test channel as a place those results could be taken from, and the issuer’s own controls should have responded to that. The control available was the one in the opening section: do not put real results into a test copy. An issuer that sent placeholders lost nothing when the return copies were taken.

Clayton’s statement also described the wider exposure the SEC itself carried. It listed three categories of data the Commission receives, including nonpublic drafts of rule filings by self-regulatory organizations and drafts of applications for exemptive relief, and nonpublic and personally identifiable information gathered in examinations and enforcement. It said the Commission faced risks of threat actors attempting to compromise the credentials of authorized EDGAR users, gain unauthorized access to filings data, place fraudulent filings on the system, and deny the public access. For a broker-dealer or investment adviser, that list is a reminder that its own examination responses, drafts and data productions also sit on a regulator’s systems, and that the data-minimization logic applies there too: send what is asked for, and no more.

Credentials as the front door

The complaint also quoted the warnings the hacker saw on EDGAR pages, including the notice that a filer must have authorized access codes, obtained through a Form ID and an SEC-issued CIK. The attack described in the complaint went around that front door rather than through it, by circumventing authentication and posing as an authorized filer. In the hindsight section we cover how the SEC changed the front door itself in 2024. For an issuer in 2019, the relevant point was narrower: EDGAR access codes were a credential like any other, worth protecting, rotating when staff or agents changed, and holding in as few hands as the filing process allowed.

The clock: the window between extraction and release

Every copy on the pre-release path is dangerous only for as long as the information is not public. The complaint showed how short a usable window can be. Its example for one issuer, identified as Issuer 2: at approximately 3:32 p.m. ET on May 19, 2016, a test filing containing positive results was uploaded to EDGAR; the hacker accessed and extracted it “only six minutes later”; a trading entity bought the stock from approximately 3:42 p.m. to 3:59 p.m. ET; the issuer released its results that day once the market had closed; and the stock rose 14 percent the next trading day. The entity closed its position by 10:46 a.m. ET on May 20, 2016, making approximately $274,663 in gross illegal profits, according to the complaint.

In another example, Issuer 1, the complaint said the test filing extracted at 1:09 p.m. ET on May 4, 2016 had been uploaded to EDGAR less than an hour earlier. The entity sold the stock short between 2:57 p.m. and 3:59 p.m. ET, and the issuer released its results once the market had closed, minutes from the last short sale. The complaint said the stock dipped 9 percent the next day.

The window is a variable the issuer sets

These examples show that the window starts when a copy with real data reaches a system the issuer does not control, and ends at publication. The issuer sets both ends. An upload of real results at midday for a release at 4:05 p.m. leaves the window open all afternoon, and the case shows that the attackers in this scheme could use a window of minutes. Moving the live submission closer to publication, and keeping real data out of tests, shrinks the window toward zero.

There is a trade-off. Testing early gives a filing team time to fix errors before a deadline. Pushing every step to the last minute increases the risk of a failed or late filing, which carries its own disclosure and market risks. The answer is to separate the two: test early with placeholders, so that format problems surface in good time, and submit the real content as late as the process safely allows. The same applies to newswire submissions. A release scheduled for a fixed time does not need to sit at the newswire with final numbers for longer than the newswire needs to process it.

Measure it

The window can be measured, and what is measured can be managed. For each earnings cycle, the filing team can record when each copy with real figures left the issuer’s systems, and to whom, and when the information was published. Over a few quarters, that record shows how long real figures wait outside the walls and where the longest waits are. It also gives the disclosure committee a concrete number to discuss rather than a general sense that the process is careful.

Timing also matters for detection

The 2015 and 2019 releases both describe trading concentrated in the gap between theft and publication. That gap is also where detection is possible. A trading pattern that recurs across issuers, in the window before scheduled releases, and only for issuers whose drafts passed through one system, is a signal. The SEC’s release credited its staff’s analysis of the trading. Issuers do not have that market-wide view. Exchanges, regulators and broker-dealers have parts of it. The next section turns to them.

The trading end: what broker-dealers could see

Information stolen from a pre-release system is worth something only if it can be traded. The complaint described how the traders did that. They used multiple brokerage accounts and entities, and accounts in the names of nominees. One entity traded through a U.S.-based brokerage account. One trader traded in his own account and in accounts in the names of at least three other people in Russia; account opening documents identified him as the “investment advisor” with exclusive trading authority for those accounts. An entity described as headquartered in Belize appeared, based on the IP addresses used to access its brokerage accounts, to be trading from Ukraine. The newswire phase, the 2015 release said, used contracts for difference and multiple accounts in a variety of names.

For broker-dealers, the EDGAR case is a reminder that their own customer onboarding and surveillance are part of the same perimeter, seen from the other side. The pre-release path ends at publication. The trading path that exploits a breach of it runs through brokerage accounts. A broker-dealer cannot prevent a newswire or a regulator from being hacked. It can be in a position to notice when its accounts trade in a pattern that looks like someone has the results in advance.

What the patterns in the complaint looked like

The complaint described several patterns that a surveillance team can think about in general terms. Trading clustered in the window before scheduled earnings releases, across issuers, in both directions, profiting on positive and negative news alike. Positions opened shortly before release and closed early the next trading day. Accounts held in other people’s names but controlled by one person. Account access from IP addresses in countries that did not match the account holder or the entity’s stated headquarters. A small entity that made a run of profitable trades and then closed its account and moved the proceeds out; the complaint said one entity’s account was closed less than two months from its last trade.

None of those patterns proves anything on its own. Together, and across issuers, they are the kind of signal the SEC said its staff used. What a broker-dealer is required to monitor, report and do about such patterns depends on its own obligations, which counsel and compliance should set. The point for this analysis is narrower: the firm’s surveillance and onboarding data are the place where a theft from someone else’s pre-release system becomes visible, and a firm that understands the pre-release path can tune its view of earnings-window trading accordingly.

The firm as an issuer’s intermediary

Broker-dealers, banks and advisers also sit on the pre-release path for their clients. Underwriters, advisers on transactions, and research and investor-relations service providers hold unreleased information. Their information barriers were designed for that role. The EDGAR case suggests one test of those barriers: whether they cover systems as well as people. A barrier that controls which bankers may see a client’s draft, but not which systems hold copies, how long those copies stay, or what the logs pointing to them reveal, is incomplete in the same way the issuers’ controls were.

A practical step is to map, for one live matter, every system that holds a copy of the client’s unreleased information: email, document management, deal rooms, collaboration tools, printing and distribution vendors, and any regulatory submission. For each, record who can reach it, how authentication works, how long the copy stays and how access is logged. The map may be longer than the barrier policy assumes. The gap between them is where the EDGAR case says to look.

Customer data and market data in the same systems

The October 2017 update, with its finding that a test filing held two people’s Social Security numbers, connects this to customer data protection. A system that stores pre-release documents for a broker-dealer or adviser can hold customer personal information too. Rules for customer information at those firms were amended in 2024, as the hindsight section describes. A program that maps pre-release systems for market-abuse reasons will find some of the same systems a customer-data program needs to map. Doing the work once serves both.

A control set for the pre-release path

The four parts of the path and the trading end translate into a short set of controls. Each is a design choice, not a statement of what any rule requires; organizations should check with counsel how each fits their own obligations.

  • Data in rehearsal copies. A written rule that test filings, proofs and previews carry placeholder financial figures and no personal data, with a quarterly sample to confirm it.
  • Copies outside the walls. A per-cycle record of every copy with real figures that leaves the firm, to whom, when, and when the information was published.
  • Intermediary contracts. Terms with newswires, filing agents and printers covering authentication, retention of drafts, staff access, logging and prompt notice of suspected exposure.
  • Filing credentials. A short, reviewed list of the people and agents who can act for the filer on EDGAR, updated when staff or agents change.
  • Incident response for market-sensitive systems. A step that lists the unreleased documents a breached system held, their publication times, and a route to the people who can review trading before publication.
  • System-level barriers. A map of the systems that hold a client’s or the firm’s own unreleased information, with owners, access rules and retention for each.
  • Surveillance at the trading end. For broker-dealers, a review of how earnings-window trading, nominee and third-party-controlled accounts, and access from unexpected locations are surfaced to compliance.

These controls divide cleanly by owner. The first four belong to the issuer’s finance, investor relations and corporate secretary functions. The fifth belongs to security and incident response, with a link to compliance. The sixth belongs to whoever owns the information barrier policy. The seventh belongs to broker-dealer compliance and surveillance. Assigning each to a named owner turns the argument of this post into a set of things that can be checked.

The record of the EDGAR case supports the first five directly. The test filings held real results and, in one case, Social Security numbers. The return copies were kept on SEC servers. Newswires held drafts for windows of minutes to days. The intrusion was detected in 2016, and the Commission learned in August 2017 that it may have provided a basis for illicit trading. The sixth and seventh follow from the same record seen from the side of firms that hold clients’ information and carry clients’ trades.

Objections, and where the argument stops

The strongest objection is that the EDGAR case was a failure of the SEC’s software, and that drawing lessons for issuers and broker-dealers blames the victims. That is fair as far as the SEC’s own systems go. Clayton’s statement identified a software vulnerability in the test filing component as the cause, and the October 2017 update set out a program of review and remediation at the agency. The argument here does not say issuers caused the breach. It says issuers controlled what was there to be taken, and that a control at the issuer would have made the breach worthless for these traders.

A second objection is cost. Placeholder testing, tighter timing and vendor oversight for publication intermediaries all add steps to a process that already runs under deadline pressure each quarter. The answer is that the steps are small compared with the risks they address, and that they can be built into the filing calendar once. Using placeholder data for a test submission costs little. Shortening the time a final release waits at a newswire costs a conversation with the vendor about processing time. The inventory of systems in the previous section is the costliest step, and it also serves other programs.

A third objection is that some exposure cannot be removed. The live filing and the newswire submission have to carry the real numbers and have to leave the issuer before publication. The argument accepts this. The aim is to make the necessary copies few, short-lived and well protected, and the unnecessary copies non-existent.

Where the argument is weakest

The weakest point is evidence of effect. The record shows what was taken and how it was traded. It does not show how many issuers sent real results in test filings, or whether any issuer changed its practice as a result, and we do not claim either. The argument is a governance argument: treat every copy of unreleased information as part of the information barrier and the security perimeter, because the record shows copies outside the issuer were taken and traded. It does not promise that any single control would have stopped every phase of this scheme.

There is also a limit to what a single organization can see. The pattern that tied these trades together was visible to a regulator with a market-wide view. An issuer sees its own releases, a broker-dealer sees its own accounts, and a newswire sees its own systems. Each holds one part of the picture. This is a reason for each to do its own part well and to escalate quickly when something looks wrong, not a reason to assume someone else will connect the dots.

With hindsight

The October 2, 2017 update had said that the EDGAR system had been undergoing modernization efforts, and listed a focused review of EDGAR, with outside consultants and an increased focus on cybersecurity, as one of its work streams. On September 27, 2024, the SEC adopted rule and form amendments, known as EDGAR Next, intended to enhance the security of EDGAR and improve filers’ access and account management. The amendments require EDGAR filers to authorize identified individuals who will be responsible for managing their accounts, and require individuals acting on behalf of filers to present individual account credentials obtained from Login.gov to access EDGAR accounts and make filings. Form ID, the application for access, was modernized. The SEC also offered optional application programming interfaces for submissions, retrieval and account management. The release said compliance with amended Form ID was required on March 24, 2025, that all amendments were effective that date, and that filers had to comply with all of them by September 15, 2025.

EDGAR Next addresses a different weakness from the one exploited in 2016. The complaint described an attack that went around authentication and posed as an authorized filer. EDGAR Next tied filing access to named individuals with their own credentials, which addresses the risk Clayton had named of threat actors compromising the credentials of authorized users. For issuers and their agents, it moved part of the pre-release path into an access model that can be governed: who is authorized, and who acts for the filer, is now something the filer records, through authorized account managers and individual credentials.

The civil case reached judgments against the two defendants who did not respond. According to an SEC litigation release dated August 5, 2021, the U.S. District Court for the District of New Jersey entered a default judgment on July 29, 2021 against the Ukrainian hacker and one of the traders, neither of whom had answered or otherwise responded to the complaint. The court permanently enjoined both from violating the antifraud provisions of Section 10(b) of the Securities Exchange Act of 1934, Rule 10b-5 and Section 17(a) of the Securities Act of 1933. It ordered them to pay penalties of $12,405,045 and $3,283,305, respectively. The release repeated the complaint’s figures: eight traders allegedly traded before at least 157 earnings releases from May to October 2016, for at least $4.1 million in illegal profits, of which the complaint attributed $1,094,435 to that trader. A default judgment shows that the court accepted the SEC’s allegations against defendants who did not contest them. It does not test the facts at trial, and we draw no more from it than the release states.

For broker-dealers, investment companies, registered investment advisers and transfer agents, the SEC adopted amendments to Regulation S-P on May 16, 2024. They require written policies and procedures for an incident response program reasonably designed to detect, respond to and recover from unauthorized access to or use of customer information, and notice to individuals whose sensitive customer information was or is reasonably likely to have been accessed or used without authorization, “as soon as practicable, but not later than 30 days” from becoming aware of the incident. Larger entities were given 18 months from Federal Register publication to comply and smaller entities 24 months, according to the release. The October 2017 finding that an EDGAR test filing held two people’s Social Security numbers is a small example of the kind of exposure these amendments address: customer personal data in a document system built for another purpose.

Three things held up. The 2019 allegations, and the judgments that followed, showed that the pre-release path is a target in its own right, as the newswire case had shown in 2015, and that the regulator’s systems are on it. The value of keeping real data out of rehearsal copies has not changed, and none of the rules since has made it unnecessary. And the separation between detecting an intrusion and understanding its market effect, visible in the gap between the 2016 detection and the August 2017 finding Clayton described, remains a design problem for any incident response plan that covers market-sensitive systems.

One thing changed in the issuers’ favor. Filing access to EDGAR is now tied to identified individuals with their own Login.gov credentials, which gives filers a record of who can act for them. Issuers and their agents should treat that record as part of the pre-release control set and review it each quarter.

This post reports only the July 2021 judgment. It does not report the outcome of the parallel criminal charges or of the SEC’s claims against the other defendants. Readers who need the full status should consult the court docket for Civil Action No. 19-cv-505 in the District of New Jersey.

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

  • EDGAR
  • material nonpublic information
  • insider trading
  • information barriers
  • SEC enforcement
  • EDGAR Next