At least 30,000 organizations across the United States — including a significant number of small businesses, towns, cities and local governments — have over the past few days been hacked by an unusually aggressive Chinese cyber espionage unit that’s focused on stealing email from victim organizations, multiple sources tell KrebsOnSecurity. The espionage group is exploiting four newly-discovered flaws in Microsoft Exchange Server email software, and has seeded hundreds of thousands of victim organizations worldwide with tools that give the attackers total, remote control over affected systems.

On March 2, Microsoft released emergency security updates to plug four security holes in Exchange Server versions 2013 through 2019 that hackers were actively using to siphon email communications from Internet-facing systems running Exchange.

In the three days since then, security experts say the same Chinese cyber espionage group has dramatically stepped up attacks on any vulnerable, unpatched Exchange servers worldwide.

In each incident, the intruders have left behind a “web shell,” an easy-to-use, password-protected hacking tool that can be accessed over the Internet from any browser. The web shell gives the attackers administrative access to the victim’s computer servers.

Speaking on condition of anonymity, two cybersecurity experts who’ve briefed U.S. national security advisors on the attack told KrebsOnSecurity the Chinese hacking group thought to be responsible has seized control over “hundreds of thousands” of Microsoft Exchange Servers worldwide — with each victim system representing approximately one organization that uses Exchange to process email.

Microsoft said the Exchange flaws are being targeted by a previously unidentified Chinese hacking crew it dubbed “Hafnium,” and said the group had been conducting targeted attacks on email systems used by a range of industry sectors, including infectious disease researchers, law firms, higher education institutions, defense contractors, policy think tanks, and NGOs.

Microsoft’s initial advisory about the Exchange flaws credited Reston, Va. based Volexity for reporting the vulnerabilities. Volexity President Steven Adair said the company first saw attackers quietly exploiting the Exchange bugs on Jan. 6, 2021, a day when most of the world was glued to television coverage of the riot at the U.S. Capitol.

But Adair said that over the past few days the hacking group has shifted into high gear, moving quickly to scan the Internet for Exchange servers that weren’t yet protected by the security updates Microsoft released Tuesday.

“We’ve worked on dozens of cases so far where web shells were put on the victim system back on Feb. 28 [before Microsoft announced its patches], all the way up to today,” Adair said. “Even if you patched the same day Microsoft published its patches, there’s still a high chance there is a web shell on your server. The truth is, if you’re running Exchange and you haven’t patched this yet, there’s a very high chance that your organization is already compromised.”

Reached for comment, Microsoft said it is working closely with the U.S. Cybersecurity & Infrastructure Security Agency (CISA), other government agencies, and security companies, to ensure it is providing the best possible guidance and mitigation for its customers.

“The best protection is to apply updates as soon as possible across all impacted systems,” a Microsoft spokesperson said in a written statement. “We continue to help customers by providing additional investigation and mitigation guidance. Impacted customers should contact our support teams for additional help and resources.”

Meanwhile, CISA has issued an emergency directive ordering all federal civilian departments and agencies running vulnerable Microsoft Exchange servers to either update the software or disconnect the products from their networks.

Adair said he’s fielded dozens of calls today from state and local government agencies that have identified the backdoors in their Exchange servers and are pleading for help. The trouble is, patching the flaws only blocks the four different ways the hackers are using to get in. But it does nothing to undo the damage that may already have been done.

White House press secretary Jen Psaki told reporters today the vulnerabilities found in Microsoft’s widely used Exchange servers were “significant,” and “could have far-reaching impacts.”

“We’re concerned that there are a large number of victims,” Psaki said.

By all accounts, rooting out these intruders is going to require an unprecedented and urgent nationwide clean-up effort. Adair and others say they’re worried that the longer it takes for victims to remove the backdoors, the more likely it is that the intruders will follow up by installing additional backdoors, and perhaps broadening the attack to include other portions of the victim’s network infrastructure.

Security researchers have published several tools for detecting vulnerable servers. One of those tools, a script from Microsoft’s Kevin Beaumont, helps companies identify exposed servers.

KrebsOnSecurity has seen portions of a victim list compiled by running such a tool, and it is not a pretty picture. The backdoor web shell is verifiably present on the networks of thousands of U.S. organizations, including banks, credit unions, non-profits, telecommunications providers, public utilities and police, fire and rescue units.

“It’s police departments, hospitals, tons of city and state governments and credit unions,” said one source who’s working closely with federal officials on the matter. “Just about everyone who’s running self-hosted Outlook Web Access and wasn’t patched as of a few days ago got hit with a zero-day attack.”

Another government cybersecurity expert who participated in a recent call with multiple stakeholders impacted by this hacking spree worries the cleanup effort required is going to be Herculean.

“On the call, many questions were from school districts or local governments that all need help,” the source said, speaking on condition they were not identified by name. “If these numbers are in the tens of thousands, how does incident response get done? There are just not enough incident response teams out there to do that quickly.”

When it released patches for the four Exchange Server flaws on Tuesday, Microsoft emphasized that the vulnerability did not affect customers running its Exchange Online service (Microsoft’s cloud-hosted email for businesses). But sources say the vast majority of the organizations victimized so far are running some form of Internet-facing Microsoft Outlook Web Access (OWA) email systems in tandem with Exchange servers internally.

“It’s a question worth asking, what’s Microsoft’s recommendation going to be?,” the government cybersecurity expert said. “They’ll say ‘Patch, but it’s better to go to the cloud.’ But how are they securing their non-cloud products? Letting them wither on the vine.”

The government cybersecurity expert said this most recent round of attacks is uncharacteristic of the kinds of nation-state level hacking typically attributed to China, which tends to be fairly focused on compromising specific strategic targets.

“Its reckless,” the source said. “It seems out of character for Chinese state actors to be this indiscriminate.”

Microsoft has said the incursions by Hafnium on vulnerable Exchange servers are in no way connected to the separate SolarWinds-related attacks, in which a suspected Russian intelligence group installed backdoors in network management software used by more than 18,000 organizations.

“We continue to see no evidence that the actor behind SolarWinds discovered or exploited any vulnerability in Microsoft products and services,” the company said.

Nevertheless, the events of the past few days may well end up far eclipsing the damage done by the SolarWinds intruders.

This is a fast-moving story, and likely will be updated multiple times throughout the day. Stay tuned.

Update, 8:27 p.m. ET: Wired cybersecurity reporter Andy Greenberg has confirmed hearing the same number of victim numbers cited in this report: “It’s massive. Absolutely massive,” one former national security official with knowledge of the investigation told WIRED. “We’re talking thousands of servers compromised per hour, globally.” Read Greenberg’s account here.

Also, the first and former director of CISA, Chris Krebs (no relation) seems to be suggesting on Twitter that the victim numbers cited here are conservative (or just outdated already):

Update 8:49 p.m.: Included a link to one of the more recommended tools for finding systems vulnerable to this attack.

Update, 10:17 p.m.: Added mention from Reuters story, which said White House officials are concerned about “a large number of victims.”

PrismHR, a company that sells technology used by other firms to help more than 80,000 small businesses manage payroll, benefits, and human resources, has suffered what appears to be an ongoing ransomware attack that is disrupting many of its services.

Hopkinton, Mass.-based PrismHR handles everything from payroll processing and human resources to health insurance and tax forms for hundreds of “professional employer organizations” (PEOs) that serve more than two million employees. The company processes more than $80 billion payroll payments annually on behalf of PEOs and their clients.

Countless small businesses turn to PEOs in part because they simplify compliance with various state payroll taxes, and because PEOs are the easiest way for small businesses to pool their resources and obtain more favorable health insurance rates for their employees.

PrismHR has not yet responded to requests for comment. But in a notice sent to its PEO partners, PrismHR said it detected suspicious activity within its networks on Feb. 28, and that it disabled access to its platform for all users in an effort to contain the security incident.

The company said the disruption has affected 200 PEO clients across the country, and that the most immediate concern is helping PEOs ensure their customers can process payrolls this week.

“The outage may extend throughout today and possibly later, with potential impact on payroll processing,” Prism explained in a template email it suggested PEO partners share with their customers. “We are committed to ensuring everyone receives their pay as timely and as accurately as possible. For this payroll period, we will use estimates from the last available payroll period. Once the software platform is back online, we will perform a reconciliation and correct any discrepancies as soon as possible.”

Jacob Cloran is co-founder of Decimal, a company that does accounting for small businesses, many of whom rely on PEOs affected by the PrismHR outage. Decimal itself uses a PEO that relies on PrismHR.

“We don’t have a good option to run our payroll this week, and the message we’ve received from our PEO doesn’t give me a lot of confidence we’ll be able to do that,” Cloran said.

Cloran said while there are other cloud-based companies that work with multiple PEOs, PrismHR is by far the largest.

“Prism is the only real option on the PEO software market,” he said. “Everyone I know who has tried any of the others ends up back at Prism. It’s the best of all bad available options.”

PrismHR did not specify what was responsible for the suspicious network activity, but their actions so far are straight out of the textbook recommendations for responding to a ransomware outbreak. A notice from the PEO working with some of Cloran’s clients stated that PrismHR was in the process of rebuilding its entire system from data backups in a new environment.

Also, the crooks behind ransomware attacks typically wait until the weekend to unleash their malware within victim organizations, knowing that most targets will be short-staffed or out of the office at this time. PrismHR said it detected the activity on Sunday.

Ransomware victims perhaps in the toughest spot include those providing cloud data hosting and software-as-service offerings, as these businesses are often unable to serve their customers while a ransomware infestation is active.

Ransomware renders any files it touches unreadable unless and until a victim pays for a digital key needed to unlock the encryption on them. Worse, it has become almost a best practice among ransomware criminal groups to steal as much data as possible from the victim organization prior to unleashing the ransom malware within a target environment.

Some of that data is often then published on dark web victim shaming sites in a bid to force the victim company into paying up. Some companies victimized by ransomware even face dual ransom demands: One for a digital key needed to unlock access to files, and a second payment in exchange for a promise not to publish all of the stolen data. Those that refuse to be extorted are told to expect that huge amounts of sensitive company data will be published online or sold on the dark web (or both).

PrismHR said in a statement to its PEO customers that while its investigation and response to the incident is ongoing, the company “is not aware of any sensitive data being breached or compromised.”

Given the volume and sensitive nature of the data PrismHR managed on behalf of PEO clients, it’s no doubt those clients and their customers are hoping that statement is accurate as well.

A company that rents out access to more than 10 million Web browsers so that clients can hide their true Internet addresses has built its network by paying browser extension makers to quietly include its code in their creations. This story examines the lopsided economics of extension development, and why installing an extension can be such a risky proposition.

Singapore-based Infatica[.]io is part of a growing industry of shadowy firms trying to woo developers who maintain popular browser extensions — desktop and mobile device software add-ons available for download from Apple, Google, Microsoft and Mozilla designed to add functionality or customization to one’s browsing experience.

Some of these extensions have garnered hundreds of thousands or even millions of users. But here’s the rub: As an extension’s user base grows, maintaining them with software updates and responding to user support requests tends to take up an inordinate amount of the author’s time. Yet extension authors have few options for earning financial compensation for their work.

So when a company comes along and offers to buy the extension — or pay the author to silently include some extra code — that proposal is frequently too good to pass up.

For its part, Infatica seeks out authors with extensions that have at least 50,000 users. An extension maker who agrees to incorporate Infatica’s computer code can earn anywhere from $15 to $45 each month for every 1,000 active users.

Infatica’s code then uses the browser of anyone who has that extension installed to route Web traffic for the company’s customers, including marketers or anyone able to afford its hefty monthly subscription charges.

The end result is when Infatica customers browse to a web site, that site thinks the traffic is coming from the Internet address tied to the extension user, not the customer’s.

Infatica prices its service based on the volume of web traffic a customer is seeking to anonymize, from $360 a month for 40 gigabytes all the way to $20,000 a month for 10,000 gigabytes of data traffic pushed through millions of residential computers.

THE ECONOMICS OF EXTENSIONS

Hao Nguyen is the developer behind ModHeader, an extension used by more than 400,000 people to test the functionality of websites by making it easier for users to modify the data shared with those sites. When Nguyen found himself spending increasing amounts of his time and money supporting the extension, he tried including ads in the program to help offset costs.

ModHeader users protested loudly against the change, and Nguyen removed the ads — which he said weren’t making him much money anyway.

“I had spent at least 10 years building this thing and had no luck monetizing it,” he told KrebsOnSecurity.

Nguyen said he ignored multiple requests from different companies offering to pay him to insert their code, mainly because the code gave those firms the ability to inject whatever they wanted into his program (and onto his users’ devices) at any time.

Then came Infatica, whose code was fairly straightforward by comparison, he said. It restricted the company to routing web requests through his users’ browsers, and did not try to access more sensitive components of the user’s browser experience, such as stored passwords and cookies, or viewing the user’s screen.

More importantly, the deal would net him at least $1,500 a month, and possibly quite a bit more.

“I gave Infatica a try but within a few days I got a lot of negative user reviews,” he said. “They didn’t like that the extension might be using their browser as a proxy for going to not so good places like porn sites.”

Again he relented, and removed the Infatica code.

A TARGET-RICH ENVIRONMENT

These days, Nguyen is focusing more of his time on chrome-stats.com, which provides detailed information on more than 150,000 extensions. The service is free for limited use, but subscribers who pay a monthly fee can get access to more resources, such as older extension versions and details about their code components.

According to chrome-stats.com, the majority of extensions — more than 100,000 of them — are effectively abandoned by their authors, or haven’t been updated in more than two years. In other words, there a great many developers who are likely to be open to someone else buying up their creation and their user base.

The vast majority of extensions are free, although a handful that have attracted a large and loyal enough following have been able to charge for their creations or for subscription services tied to the extension. But last year, Google announced it was shutting down paid Chrome extensions offered on its Chrome Web Store.

Nguyen said this will only exacerbate the problem of frustrated developers turning to offers from dodgy marketing firms.

“It’s a really tough marketplace for extension developers to be able to monetize and get reward for maintaining their extensions,” he said. “There are tons of small developers who haven’t been able to do anything with their extensions. That’s why some of them will go into shady integration or sell the extension for some money and just be done with it.”

WHO IS INFATICA?

It is unclear how many extensions currently incorporate Infatica’s code. KrebsOnSecurity searched for extensions that invoke several domains tied to Infatica’s Web proxy service (e.g., extendbalanc[.]org, ipv4v6[.]info). This research was conducted using Nguyen’s site and crxcavator.io, a similar extension research site owned by networking giant Cisco Systems.

Those searches revealed that Infatica’s code has been associated with at least three dozen extensions over the past few years, including several that had more than 100,000 users. One of those is Video Downloader Plus, which at one point claimed nearly 1.4 million active users.

The founder and director of Infatica — a resident of Biysk, Russia named Vladimir Fomenko — did not respond to multiple requests for comment.

Fomenko is the sole director of the iNinja VPN, another service that obfuscates the true Internet address of its more than 400,000 users. It stands to reason that iNinja VPN also is not only offering its customers a way to obfuscate their Internet address, but is actively using those same systems to route traffic for other customers: A Chrome browser plugin and ad blocker by the same name whose code includes Infatica’s “extenbalanc” domain has 400,000 users.

That would put Infatica in line with the activities of another major controversial VPN/proxy provider: Luminati, a.k.a. “HolaVPN.” In 2015, security researchers discovered that users of the HolaVPN browser extension were being used to funnel Web traffic for other people. Indeed, in the screenshot above, Infatica’s marketing team can be seen comparing its business model to that of HolaVPN.

Fomenko has appeared in two previous KrebsOnSecurity stories; both concerned King Servers (a.k.a. “Hosting Solution Ltd.“), a hosting company he has operated for years which caters mostly to adult websites.

In 2016, hackers suspected of working for Russian state security services compromised databases for election systems in Arizona and Illinois. Six of the eight Internet addresses identified by the FBI as sources of the attack traced back to King Servers. In an interview with The New York Times several months later, Fomenko flatly denied having any ties to the hacking.

According to the Russian daily Novaya Gazeta, revelations about the 2016 hacking incident’s ties to King Servers led to treason charges against Sergey Mikhaylov, the former deputy chief of Russia’s top anti-cybercrime unit.

Russian authorities charged that Mikhaylov had tipped off the FBI to information about Fomenko and King Servers. In 2019, Mikhaylov was convicted and sentenced to 22 years in a penal colony.

BE SPARING IN TRUSTING EXTENSIONS

Browser extensions — however useful or fun they may seem when you install them — typically have a great deal of power and can effectively read and/or write all data in your browsing sessions. The powers granted to each extension are roughly spelled out in its “manifest,” basically a description of what it will be able to access once you incorporate it into your browser.

According to Nguyen’s chrome-stats.com, about a third of all extensions for Chrome — by far the most widely-used Web browser — require no special permissions. But the remainder require the user to place a good deal of trust in the extension’s author. For example, approximately 30 percent can view all of your data on all or specific websites, or index your open tabs and browsing activity.

More than 68,000 Chrome extensions allow the execution of arbitrary code in the context of webpages, effectively allowing the extension to alter the appearance and functionality of specific sites.

I hope it’s obvious by this point, but readers should be extremely cautious about installing extensions — sticking mainly to those that are actively supported and respond to user concerns.

Personally, I do not make much use of browser extensions. In almost every case I’ve considered installing one I’ve been sufficiently spooked by the permissions requested that I ultimately decided it wasn’t worth the risk, given that any extension can go rogue at the whims of its author.

If you’re the type of person who uses multiple extensions, it may be wise to adopt a risk-based approach going forward. Given the high stakes that typically come with installing an extension, consider carefully whether having the extension is truly worth it. This applies equally to plug-ins designed for Web site content management systems like WordPress and Joomla.

Do not agree to update an extension if it suddenly requests more permissions than a previous version. This should be a giant red flag that something is not right. If this happens with an extension you trust, you’d be well advised to remove it entirely.

Also, never download and install an extension just because some Web site says you need it to view some type of content. Doing so is almost always a high-risk proposition. Here, Rule #1 from KrebsOnSecurity’s Three Rules of Online Safety comes into play: “If you didn’t go looking for it, don’t install it.” Finally, in the event you do wish to install something, make sure you’re getting it directly from the entity that produced the software.

Google Chrome users can see any extensions they have installed by clicking the three dots to the right of the address bar, selecting “More tools” in the resulting drop-down menu, then “Extensions.” In Firefox, click the three horizontal bars next to the address bar and select “Add-ons,” then click the “Extensions” link on the resulting page to view any installed extensions.

Easily the most sophisticated skimming devices made for hacking terminals at retail self-checkout lanes are a new breed of PIN pad overlay combined with a flexible, paper-thin device that fits inside the terminal’s chip reader slot. What enables these skimmers to be so slim? They draw their power from the low-voltage current that gets triggered when a chip-based card is inserted. As a result, they do not require external batteries, and can remain in operation indefinitely.

The overlay skimming device pictured above consists of two main components. The one on top is a regular PIN pad overlay designed to record keypresses when a customer enters their debit card PIN. The overlay includes a microcontroller and a small data storage unit (bottom left).

The second component, which is wired to the overlay skimmer, is a flexible card skimmer (often called a “shimmer”) that gets fed into the mouth of the chip card acceptance slot. You’ll notice neither device contains a battery, because there simply isn’t enough space to accommodate one.

Virtually all payment card terminals at self-checkout lanes now accept (if not also require) cards with a chip to be inserted into the machine. When a chip card is inserted, the terminal reads the data stored on the smart card by sending an electric current through the chip.

Incredibly, this skimming apparatus is able to siphon a small amount of that power (a few milliamps) to record any data transmitted by the payment terminal transaction and PIN pad presses. When the terminal is no longer in use, the skimming device remains dormant.

The skimmer pictured above does not stick out of the payment terminal at all when it’s been seated properly inside the machine. Here’s what the fake PIN pad overlay and card skimmer looks like when fully inserted into the card acceptance slot and viewed head-on:

Would you detect an overlay skimmer like this? Here’s what it looks like when attached to a customer-facing payment terminal:

REALLY SMART CARDS

The fraud investigators I spoke with about this device (who did so on condition of anonymity) said initially they couldn’t figure out how the thieves who plant these devices go about retrieving the stolen data from the skimmer. Normally, overlay skimmers relay this data wirelessly using a built-in Bluetooth circuit board. But that also requires the device to have a substantial internal power supply, such as a somewhat bulky cell phone battery.

The investigators surmised that the crooks would retrieve the stolen data by periodically revisiting the compromised terminals with a specialized smart card that — when inserted — instructs the skimmer to dump all of the saved information onto the card. And indeed, this is exactly what investigators ultimately found was the case.

“Originally it was just speculation,” the source told KrebsOnSecurity. “But a [compromised] merchant found a couple of ‘white’ smartcards with no markings on them [that] were left at one of their stores. They informed us that they had a lab validate that this is how it worked.”

Some readers might reasonably be asking why it would be the case that the card acceptance slot on any chip-based payment terminal would be tall enough to accommodate both a chip card and a flexible skimming device such as this.

The answer, as with many aspects of security systems that decrease in effectiveness over time, has to do with allowances made for purposes of backward compatibility. Most modern chip-based cards are significantly thinner than the average payment card was just a few years ago, but the design specifications for these terminals state that they must be able to allow the use of older, taller cards — such as those that still include embossing (raised numbers and letters). Embossing is a practically stone-age throwback to the way credit cards were originally read, through the use of manual “knuckle-buster” card imprint machines and carbon-copy paper.

“The bad guys are taking advantage of that, because most smart cards are way thinner than the specs for these machines require,” the source explained. “In fact, these slots are so tall that you could fit two cards in there.”

IT’S ALL BACKWARDS

Backward compatibility is a major theme in enabling many types of card skimming, including devices made to compromise automated teller machines (ATMs). Virtually all chip-based cards (at least those issued in the United States) still have much of the same data that’s stored in the chip encoded on a magnetic stripe on the back of the card. This dual functionality also allows cardholders to swipe the stripe if for some reason the card’s chip or a merchant’s smartcard-enabled terminal has malfunctioned.

Chip-based credit and debit cards are designed to make it infeasible for skimming devices or malware to clone your card when you pay for something by dipping the chip instead of swiping the stripe. But thieves are adept at exploiting weaknesses in how certain financial institutions have implemented the technology to sidestep key chip card security features and effectively create usable, counterfeit cards.

Many people believe that skimmers are mainly a problem in the United States, where some ATMs still do not require more secure chip-based cards that are far more expensive and difficult for thieves to clone. However, it’s precisely because some U.S. ATMs lack this security requirement that skimming remains so prevalent in other parts of the world.

Mainly for reasons of backward compatibility to accommodate American tourists, a great number of ATMs outside the U.S. allow non-chip-based cards to be inserted into the cash machine. What’s more, many chip-based cards issued by American and European banks alike still have cardholder data encoded on a magnetic stripe in addition to the chip.

When thieves skim non-U.S. ATMs, they generally sell the stolen card and PIN data to fraudsters in Asia and North America. Those fraudsters in turn will encode the card data onto counterfeit cards and withdraw cash at older ATMs here in the United States and elsewhere.

Interestingly, even after most U.S. banks put in place fully chip-capable ATMs, the magnetic stripe will still be needed because it’s an integral part of the way ATMs work: Most ATMs in use today require a magnetic stripe for the card to be accepted into the machine. The main reason for this is to ensure that customers are putting the card into the slot correctly, as embossed letters and numbers running across odd spots in the card reader can take their toll on the machines over time.

And there are the tens of thousands of fuel pumps here in the United States that still allow chip-based card accounts to be swiped. The fuel pump industry has for years won delay after delay in implementing more secure payment requirements for cards (primarily by flexing their ability to favor their own fuel-branded cards, which largely bypass the major credit card networks).

Unsurprisingly, the past two decades have seen the emergence of organized gas theft gangs that take full advantage of the single weakest area of card security in the United States. These thieves use cloned cards to steal hundreds of gallons of gas at multiple filling stations. The gas is pumped into hollowed-out trucks and vans, which ferry the fuel to a giant tanker truck. The criminals then sell and deliver the gas at cut rate prices to shady and complicit fuel station owners and truck stops.

A great many people use debit cards for everyday purchases, but I’ve never been interested in assuming the added risk and pay for everything with cash or a credit card. Armed with your PIN and debit card data, thieves can clone the card and pull money out of your account at an ATM. Having your checking account emptied of cash while your bank sorts out the situation can be a huge hassle and create secondary problems (bounced checks, for instance).

The next skimmer post here will examine an inexpensive and ingenious analog device that helps retail workers quickly check whether their payment terminals have been tampered with by bad guys.

The U.S. Justice Department today unsealed indictments against three men accused of working with the North Korean regime to carry out some of the most damaging cybercrime attacks over the past decade, including the 2014 hack of Sony Pictures, the global WannaCry ransomware contagion of 2017, and the theft of roughly $200 million and attempted theft of more than $1.2 billion from banks and other victims worldwide.

Investigators with the DOJ, U.S. Secret Service and Department of Homeland Security told reporters on Wednesday the trio’s activities involved extortion, phishing, direct attacks on financial institutions and ATM networks, as well as malicious applications that masqueraded as software tools to help people manage their cryptocurrency holdings.

Prosecutors say the hackers were part of an effort to circumvent ongoing international financial sanctions against the North Korean regime. The group is thought to be responsible for the attempted theft of approximately $1.2 billion, although it’s unclear how much of that was actually stolen.

Confirmed thefts attributed to the group include the 2016 hacking of the SWIFT payment system for Bangladesh Bank, which netted thieves $81 million; $6.1 million in a 2018 ATM cash out scheme targeting a Pakistani bank; and a total of $112 million in virtual currencies stolen between 2017 and 2020 from cryptocurrency companies in Slovenia, Indonesia and New York.

“The scope of the criminal conduct by the North Korean hackers was extensive and longrunning, and the range of crimes they have committed is staggering,” said Acting U.S. Attorney Tracy L. Wilkison for the Central District of California. “The conduct detailed in the indictment are the acts of a criminal nation-state that has stopped at nothing to extract revenge and obtain money to prop up its regime.”

The indictments name Jon Chang Hyok (a.k.a “Alex/Quan Jiang”), Kim Il (a.k.a. “Julien Kim”/”Tony Walker”), and Park Jin Hyok (a.k.a. Pak Jin Hek/Pak Kwang Jin). U.S. prosecutors say the men were members of the Reconnaissance General Bureau (RGB), an intelligence division of the Democratic People’s Republic of Korea (DPRK) that manages the state’s clandestine operations.

The Justice Department says those indicted were members of a DPRK-sponsored cybercrime group variously identified by the security community as the Lazarus Group and Advanced Persistent Threat 38 (APT 38). The government alleges the men reside in North Korea but were frequently stationed by the DPRK in other countries, including China and Russia.

Park was previously charged in 2018 in connection with the WannaCry and Sony Pictures attacks. But today’s indictments expanded the range of crimes attributed to Park and his alleged co-conspirators, including cryptocurrency thefts, phony cryptocurrency investment schemes and apps, and efforts to launder the proceeds of their crimes.

Prosecutors in California also today unsealed an indictment against Ghaleb Alaumary, a 37-year-old from Mississauga, Ontario who pleaded guilty in November 2020 to charges of laundering tens of millions of dollars stolen by the DPRK hackers.

The accused allegedly developed and marketed a series of cryptocurrency applications that were advertised as tools to help people manage their crypto holdings. In reality, prosecutors say, the programs were malware or downloaded malware after the applications were installed.

A joint cyber advisory from the FBI, the Treasury and DHS’s Cybersecurity and Infrastructure Agency (CISA) delves deeper into these backdoored cryptocurrency apps, a family of malware activity referred to as “AppleJeus. “Hidden Cobra” is the collective handle assigned to the hackers behind the AppleJeus malware.

“In most instances, the malicious application—seen on both Windows and Mac operating systems—appears to be from a legitimate cryptocurrency trading company, thus fooling individuals into downloading it as a third-party application from a website that seems legitimate,” the advisory reads. “In addition to infecting victims through legitimate-looking websites, HIDDEN COBRA actors also use phishing, social networking, and social engineering techniques to lure users into downloading the malware.”

The alert notes that these apps have been posing as cryptocurrency trading platforms since 2018, and have been tied to cryptocurrency thefts in more than 30 countries.

For example, the DOJ indictments say these apps were involved in stealing $11.8 million in August 2020 from a financial services company based in New York. Warrants obtained by the government allowed the FBI to seize roughly $1.9 million from two different cryptocurrency exchanges used by the hackers, money that investigators say will be returned to the New York financial services firm.

Other moneymaking and laundering schemes attributed to the North Korean hackers include the development and marketing of an initial coin offering (ICO) in 2017 called Marine Chain Token.

That blockchain-based cryptocurrency offering promised early investors the ability to purchase “fractional ownership in marine shipping vessels,” which the government says was just another way for the North Korean government to “secretly obtain funds from investors, control interests in marine shipping vessels, and evade U.S. sanctions.”

A copy of the indictments is available here (PDF).