Threat Researcher at Gen

Our team invests countless hours in the exploration, discovery and investigation of emerging trends such as AI and futuristic technology. Stay informed about the evolving landscape of cyberthreats through our diligent efforts and insights.

Research

<h2>Introduction&nbsp;</h2>Infostealers are constantly evolving, and so are the techniques they use to bypass <a href="https://security.googleblog.com/2024/07/improving-security-of-chrome-cookies-on.html" target="_blank" rel="noreferrer noopener">Application-Bound Encryption (ABE)</a>. In recent weeks, Vidar has been among the most actively developed stealers and, apart from <a href="https://x.com/GenThreatLabs/status/2064836539952361976" target="_blank" rel="noreferrer noopener">multiple updates to its string obfuscation</a> and <a href="https://x.com/GenThreatLabs/status/2061462524604649840" target="_blank" rel="noreferrer noopener">a reworked approach to protecting its configuration</a>, it has also introduced a novel technique for bypassing ABE. And while there have been many other changes in Vidar lately, with new versions dropping every week, in this blog post we focus solely on the ABE bypass and its technical aspects.&nbsp; &nbsp;<h2>The ABE Bypass&nbsp;</h2>From a high-level perspective, Vidar’s approach to bypassing ABE is somewhat similar to that of Remus/Lumma, in that both try to extract the <code>v20_master_key</code> directly from the browser’s memory. What sets them apart, however, is how they achieve it. Since we’ve already covered the inner workings of ABE in earlier posts (<a href="https://www.gendigital.com/blog/insights/research/remus-64bit-variant-of-lumma-stealer" target="_blank" rel="noreferrer noopener">Remus</a>, <a href="https://www.gendigital.com/blog/insights/research/voidstealer-abe-bypass" target="_blank" rel="noreferrer noopener">VoidStealer</a>), we don’t revisit them here and focus instead on the bypass itself. For readers interested in the broader background, we encourage checking those earlier posts.&nbsp;&nbsp;In short, the end goal of infostealers is to obtain the <code>v20_master_key</code>, as it alone is sufficient for decrypting any ABE-protected data tied to a specific application. This key can be obtained in one of two ways: either from disk, where it is protected by two layers of <code>CryptProtectData</code> (one applied in the context of the logged-in user and one as <code>NT AUTHORITY\SYSTEM</code>), or from the browser’s memory – the direction Vidar took.&nbsp;In memory, however, the key is kept protected with <code>CryptProtectMemory</code> using the <code>CRYPTPROTECTMEMORY_SAME_PROCESS</code> flag. This ensures that even if an attacker manages to read the encrypted form of the key from memory, they cannot simply decrypt it (they would need to invoke the complementary <code>CryptUnprotectMemory</code> function directly from within the browser process). That leaves attackers with two problems to solve: how to find the encrypted key in memory, and how to get it decrypted.&nbsp;<h2>Locating the v20_master_key in Browser’s Memory&nbsp;</h2>Vidar’s approach to locating the encrypted <code>v20_master_key</code> in memory involves scanning the entire memory of a running browser for a specific pattern. But that first requires a browser process to scan.&nbsp;When the target browser is already running, Vidar reuses the existing process. However, it doesn’t read its memory directly. Instead, it creates a fork of it using <code>NtCreateProcessEx</code> with <code>SectionHandle</code> set to <code>NULL</code>, instructing the kernel to clone the address space of the process specified by the <code>ParentProcess</code> as a copy-on-write mapping. The resulting process has no threads and is never resumed – it exists purely as a static memory snapshot.&nbsp;&nbsp;To perform the fork, Vidar first opens a handle to the target browser via <code>OpenProcess</code> with <code>PROCESS_QUERY_INFORMATION | PROCESS_CREATE_PROCESS | PROCESS_VM_READ</code>, falling back to <code>PROCESS_CREATE_PROCESS</code> if the initial call fails. It then passes this handle as the <code>ParentProcess</code> argument to <code>NtCreateProcessEx</code>, requesting <code>PROCESS_ALL_ACCESS</code> on the resulting child. If the fork fails, Vidar falls back to scanning the live browser process directly.&nbsp;

If no target browser process is found running, Vidar creates a new isolated desktop via <code>CreateDesktopA</code>, assigns it a name of the form <code>v20_%d</code> (where <code>%d</code> is replaced with the browser’s index in Vidar’s configuration), and launches the target browser in there. In the case of Vidar version 2.1, the browser is launched with the command line <code>--no-first-run --disable-gpu about:blank</code>, although the exact command line varies across versions. Since modern browsers use a multi-process architecture, there are typically multiple processes matching a given browser. Vidar enumerates all of them (up to 64) and applies the fork-and-scan procedure to each one independently.&nbsp;

With a handle to the forked process in hand, Vidar proceeds to enumerate its virtual memory regions using <code>NtQueryVirtualMemory</code>. It iterates through the entire browser’s address space, collecting regions that match a specific profile: committed (<code>MEM_COMMIT</code>), private (<code>MEM_PRIVATE</code>), and either readable (<code>PAGE_READONLY</code>) or read-write (<code>PAGE_READWRITE</code>). Vidar records up to 4096 memory regions matching these requirements, along with their base addresses and sizes. The actual pattern scanning is then performed in parallel with up to 64 worker threads, depending on the system.&nbsp;&nbsp;

Each thread scans its assigned regions using <code>NtReadVirtualMemory</code>, searching for a 32-byte pattern. When a match is found, the thread reads 8 bytes at offset +32 from the match, interprets them as a pointer, and validates that it points to committed, private, and readable memory (the expected memory state for the encrypted <code>v20_master_key</code>).&nbsp;&nbsp;

This naturally raises the question of what pattern Vidar searches for. The answer is a 32-byte signature: <code>76 32 30 00 ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? ?? 03 00 00 00 00 01 ?? ?? ??</code>, where <code>??</code> denotes a wildcard (Vidar uses the byte <code>AA</code> as the wildcard marker in its internal representation). Notably, the first four bytes correspond to the ASCII <code>v20\x00</code>, which is a common prefix used for ABE-related data.&nbsp;&nbsp;

We assess with high confidence that this pattern is intended to match a node structure within Chromium’s <code>Encryptor::KeyRing</code>, the structure that holds the keys used by the <code>Encryptor</code> class. Now, this part is admittedly a bit tricky as it relies on several assumptions we made about the internal memory layout, as well as on implementation-specific representations of various C++ standard library types <code>(std::map</code>, <code>std::string</code>, <code>std::vector</code>, <code>std::optional</code>), which is why we frame this as a high-confidence rather than a certainty.&nbsp;&nbsp;Our reasoning is as follows: <code>KeyRing</code> is <a href="https://github.com/chromium/chromium/blob/cf5d502bcaedb22927e3829a18be1ff6d8b6f598/components/os_crypt/async/common/encryptor.h#L132" target="_blank" rel="noreferrer noopener">defined</a> as <code>std::map&lt;/*tag=*/std::string, std::optional&lt;Key&gt;&gt;</code>, and <code>std::map</code> is typically implemented as a red-black tree, where each tree node stores three pointers (left child, parent, and right child) followed by the key-value pair. In this case, the pair is in the form of <code>std::pair&lt;/*tag=*/std::string, std::optional&lt;Key&gt;&gt;</code>, meaning that the <code>std::string</code> and the <code>Key</code> object might be laid out consecutively in memory within each node. For a <code>v20_master_key</code> entry, the tag corresponds to the string <code>v20</code>, which at 4 bytes including the null terminator fits within the small string optimization (SSO) buffer and is therefore stored inline rather than heap-allocated. The <code>Key</code> class that immediately follows <a href="https://github.com/chromium/chromium/blob/cf5d502bcaedb22927e3829a18be1ff6d8b6f598/components/os_crypt/async/common/encryptor.h#L93-L98" target="_blank" rel="noreferrer noopener">contains three member variables</a>: <code>std::optional&lt;mojom::Algorithm&gt; algorithm_</code>, <code>std::vector&lt;uint8_t&gt; key_</code>, and <code>bool encrypted_ = false</code>.&nbsp;Putting it all together and applying it to a concrete example from a browser memory dump, we get the following layout:&nbsp;

Please note that while the exact layout may contain minor inaccuracies on our part, the overall structure should be roughly correct. For easier orientation, the following diagram illustrates how the individual structures are nested within each other:&nbsp;

At this point, it should be clear how the scan pattern maps onto the searched structure. The 32-byte match covers the <code>std::string</code> tag <code>v20\x00</code> and the beginning of the <code>Key</code> object up to and including <code>algorithm_</code>. The pointer at offset +32 from the start of the match then lands precisely on <code>key_.begin</code>, the first member of <code>std::vector&lt;uint8_t&gt;</code> that holds the encrypted <code>v20_master_key</code>.&nbsp;Matches from all threads are collected into a shared buffer and subjected to a majority-voting scheme: candidates consisting mostly of zeroes are discarded, the rest are pairwise-compared, and each is scored by the number of identical copies found. This voting is performed independently for each browser process, producing one best candidate per process.&nbsp;<h2>Decrypting the v20_master_key via APC Injections&nbsp;</h2>For each browser process that produced a match, Vidar now holds a candidate address pointing to a potential encrypted <code>v20_master_key</code>. However, as already noted, since the key is protected with <code>CryptProtectMemory</code> using <code>CRYPTPROTECTMEMORY_SAME_PROCESS</code>, it can only be decrypted by invoking <code>CryptUnprotectMemory</code> from within the browser process itself. Vidar solves this by injecting an Asynchronous Procedure Call (APC) into the live browser process.&nbsp;Vidar supports two APC injection methods, selected based on whether ESET or Bitdefender is detected on the system. If either of these products is present, Vidar uses the first method, which it refers to as “classic” (based on decrypted strings) – it creates a suspended thread in the target browser via <code>CreateRemoteThread</code> (with <code>NtTestAlert</code> as the thread start address and the <code>CREATE_SUSPENDED</code> flag), queues an APC using <code>NtQueueApcThread</code>, resumes the thread, and then waits for it to complete via <code>WaitForSingleObject</code> with a randomized timeout. When the thread resumes, it calls <code>NtTestAlert</code>, which flushes the APC queue and executes the injected routine.&nbsp;

If neither product is detected, it uses the “special” method – it finds an existing thread in the target browser using a combination of <code>CreateToolhelp32Snapshot</code> and <code>Thread32First/Thread32Next</code>, opens it via <code>NtOpenThread</code>, and calls <code>NtQueueApcThreadEx2</code> with the <code>QUEUE_USER_APC_FLAGS_SPECIAL_USER_APC</code> flag. Unlike regular APCs, special user APCs execute immediately. Therefore, the target thread does not need to be in an alertable wait state.&nbsp;

In both cases, the APC routine to be executed is <code>CryptUnprotectMemory</code>, called with the following three arguments: the address of the encrypted <code>v20_master_key</code> candidate, its size (32 bytes), and <code>CRYPTPROTECTMEMORY_SAME_PROCESS</code> for the <code>flags</code> parameter.&nbsp;&nbsp;Since <code>CryptUnprotectMemory</code> decrypts the data in place, once the APC executes, the key is decrypted directly in the live browser's memory. To read it, Vidar creates a second fork of the browser process and reads the key from the same address using <code>NtReadVirtualMemory</code>. It then compares the result against a snapshot of the encrypted key taken from the first fork before the APC call – if the data is unchanged, decryption failed, and Vidar moves on to the next browser process’s candidate.&nbsp;If the data differs, Vidar uses the same second fork to perform an additional verification: it scans the forked process’s memory for the byte sequence <code>v20</code> and, treating each match as a potentially ABE-encrypted entry (cookie, password), attempts AES-256-GCM decryption with the key candidate using <code>BCryptDecrypt</code> (successful GCM tag verification confirms that the key is correct).&nbsp;Whenever the data has changed (indicating that decryption of the key took place), Vidar re-encrypts the key in the browser’s memory by injecting another APC with <code>CryptProtectMemory</code> as the routine, preserving the browser’s state. If no candidate produced a successful decryption (i.e., the data remained unchanged after every APC call), Vidar takes a more aggressive approach: it terminates all browser instances, restarts the browser, and repeats the entire scan-decrypt-verify cycle.&nbsp;By abusing APC injections instead of the more traditional approach of injecting code via <code>NtWriteVirtualMemory</code> that performs the decryption, Vidar trades one suspicious behavior for another. APC injections are arguably less common and may thus slip past some detection systems, though they are still well-known techniques monitored by AV and EDR solutions. But that is the nature of ABE bypasses – each has its own advantages and disadvantages, and that’s exactly why infostealers keep inventing new ones. &nbsp;<h2>Indicator of Compromise (IoC)&nbsp;</h2>459daa809751e73f60fbbe4384a7d1653c36bb06945e4eb3635270924241100a (Vidar 2.1)&nbsp;

A Technical Walkthrough of How Vidar Defeats Application-Bound Encryption

Inside Vidar’s ABE Bypass: From Memory Scanning to APC Injections

Security Evangelist at Gen

You have a summer flight booked. You have seen the headlines about jet fuel shortages, cancelled routes and higher travel costs. Then a text message arrives: your flight has been affected, and you need to click a link to confirm your refund or rebook before the offer expires.&nbsp;That message feels plausible because the situation around it is real. In April, <a href="https://www.iata.org/en/pressroom/2026-releases/2026-04-17-01/" target="_blank" rel="noreferrer noopener">IATA warned</a> that Europe could start seeing flight cancellations by the end of May if jet fuel shortages materialized. Airlines and governments have since been trying to reassure travelers, while also publishing guidance on passenger rights, refunds and re-routing.&nbsp;&nbsp;That is exactly the kind of confusion scammers like. They do not need to cause the disruption, they only need to arrive at the right moment, with the right brand logo and the right sense of urgency.&nbsp;<h4>Why travel disruption works so well as a scam hook&nbsp;</h4>Most scam messages fail because they arrive out of context. A random refund email from an airline you never use is easy to ignore. A refund message from the airline you are flying with next week, during a period of public travel disruption, is different.&nbsp;The scammer’s job is to make you think: “This could be real.”&nbsp;That can happen in several ways. You might receive a fake cancellation email that links to a refund form. You might search for an airline’s customer service number and land on a fake support page. You might post publicly on social media asking an airline for help, then receive a reply from an account that looks official. Or you might get a WhatsApp message that knows your hotel name, travel dates or booking reference.&nbsp;Each version uses the same pressure point: you are worried about losing money, losing your flight or being stranded.&nbsp;Criminals are already using disruption this way. In March, <a href="https://www.theguardian.com/money/2026/mar/15/travel-scam-airline-accounts-fake-refunds-iran-war-flight-disruption" target="_blank" rel="noreferrer noopener">The Guardian reported</a> that fake airline support accounts were targeting travelers affected by Middle East flight disruption. Some victims were looking for help or refunds, then received replies from accounts pretending to represent airlines. The scam moved into direct messages, where victims were asked for details and sent links that led to money being taken instead of refunded.&nbsp;&nbsp;The jet fuel story gives scammers another pretext. A message can claim that your flight was cancelled, your fare changed, your fuel surcharge failed, or your refund must be claimed within a few hours. None of those claims needs to be true to be effective.&nbsp;<h4>The scam may start before the message arrives&nbsp;</h4>A fake travel message becomes much harder to spot when it includes real information.&nbsp;This is the idea behind <a href="https://www.gendigital.com/blog/insights/research/reservation-hijack-scam" target="_blank" rel="noreferrer noopener">Reservation Hijack scams</a>, a form of targeted phishing where cybercriminals use real booking details to make a message look legitimate. In some cases, attackers may use information exposed through breaches or compromised travel systems. In other cases, they collect details from emails, social media posts, previous leaks or fake customer service interactions. Gen has previously described Reservation Hijack scams as scams that use real hotel booking details to make fraudulent payment or verification requests feel authentic.&nbsp;&nbsp;That distinction matters. A scam message does not need your credit card number to fool you. Knowing your name, your destination, your travel dates and your booking platform may be enough.&nbsp;<h4>What these scams can look like&nbsp;</h4>A travel disruption scam usually does not begin with a strange offer. It begins with something boring and administrative. That is part of the trick.&nbsp;You may see messages like:&nbsp;“Your flight has been cancelled due to fuel supply issues. Claim your refund here.”&nbsp;“Your rebooking is pending. Confirm passenger details within two hours.”&nbsp;“Additional airport tax or fuel surcharge is required to keep your booking.”&nbsp;“Your hotel reservation will be cancelled unless payment is verified.”&nbsp;“Due to high call volume, please continue this request on WhatsApp.”&nbsp;The exact wording changes, but the structure is familiar. There is a problem, a deadline and a link. The message may also move you away from the official channel into SMS, WhatsApp, Telegram, direct messages or a phone call. Once that happens, the scammer controls the conversation.&nbsp;There is another detail worth watching: some messages do not ask for money immediately. They start by asking for your booking reference, last name, phone number or one-time password. That can still be dangerous. Those details can help scammers access a booking, impersonate you to customer support, reset an account or make the next message more convincing.&nbsp;<h4>Your real passenger rights can be used against you&nbsp;</h4>If an airline cancels your flight, you may have rights to a refund, re-routing or assistance, depending on where you are travelling and the rules that apply. The European Commission said passengers affected by cancellations continue to have rights to reimbursement, re-routing or return, airport assistance and compensation for last-minute cancellations in relevant circumstances. It also said airlines cannot retroactively add fuel surcharges to already published final ticket prices.&nbsp;&nbsp;That information is useful for travelers, but it is also useful for scammers.&nbsp;A fake refund message becomes more believable when people already know refunds are possible. A fake rebooking link becomes more believable when airlines are genuinely changing schedules. A fake “fuel surcharge” demand becomes more dangerous when fuel prices are in the news, even though legitimate retroactive charges may not be allowed in many cases.&nbsp;This is why the safest route is not to judge the message only by whether the story sounds plausible. During travel disruption, many scam stories will sound plausible. Judge the channel instead.&nbsp;<h4>How to check before you click&nbsp;</h4>Start from the airline or booking platform’s official app or website. Do not use the link in the message. Open the app yourself, type the website address yourself, or use the contact details listed on your original booking confirmation.&nbsp;Check whether the same cancellation, refund or payment request appears inside your account. If the message is real, the airline or hotel should be able to confirm it through an official channel.&nbsp;Be careful with social media support. If an account replies to your public complaint, check whether it is linked from the airline’s official website. Look at the handle carefully. Fake support accounts often copy logos and names, but have small differences in spelling, low follower counts or recently created profiles. The Guardian reported that fraudsters often scan social media for passengers trying to contact legitimate airline accounts and reply to them first, which makes the scam harder to spot.&nbsp;&nbsp;Do not share one-time passwords, payment card details or login credentials with anyone who contacts you unexpectedly. A real airline or hotel does not need your banking OTP to send you a refund.&nbsp;Be especially suspicious of pressure. “You have 30 minutes to claim your refund” is not customer service. It is a tactic.&nbsp;<h4>What to do if you already clicked&nbsp;</h4>If you entered payment details, contact your bank or card provider immediately. Ask whether the transaction can be blocked or reversed, and request a new card if needed.&nbsp;If you shared login details for an airline, hotel or booking platform, change the password from the official site or app. Use a unique password, not one you use elsewhere. Turn on two-factor authentication where it is available.&nbsp;If you shared a one-time password or authentication code, assume the account may have been accessed. Check your booking details, payment methods and recent account activity.&nbsp;Report the scam to the airline, hotel or booking platform being impersonated. If the scam came through social media, report the account as impersonation. If money was taken, report it to the relevant fraud authority in your country.&nbsp;<h4>The real lesson for summer travel&nbsp;</h4>Travel disruption changes how people behave. They search for support numbers in a hurry. They post publicly because customer service queues are long. They accept strange instructions because they are worried about losing a flight or refund.&nbsp;Scammers build around that behavior.&nbsp;The safest assumption is simple: if a message about your flight, hotel or refund creates urgency and gives you a link, pause. Go back to the official source. A real cancellation will still be there when you check through the airline or booking platform. A scammer needs you to act before you do.&nbsp;

Travel disruption gives scammers the one thing they need most: a believable reason to rush you

Your flight was cancelled. Is the refund message real? 

A fake job interview page is easier to trust than a fake ticket shop.In samples reviewed by Gen threat researchers, scammers used branded hiring pages that looked like ordinary recruitment flows: a company logo, a recruiter profile, a 30-minute meeting, and a button to continue with Google or Facebook. There was no attachment to open and no software to install. The credential theft attempt sat inside a step many job applicants already expect: signing in to schedule a call.FIFA is an obvious lure ahead of the 2026 World Cup. On May 22, ESET reported fake FIFA-themed websites that copied ticket and merchandise flows to steal money and personal data. Our researchers later found another FIFA-themed domain, hxxps://fifa-mg[.]shop, which followed a naming pattern not mentioned in that report. Domains like these can be short-lived, often appearing for hours or days before they are blocked, abandoned, or taken down.Our researchers also found a different kind of FIFA lure: hxxps://fifahiring[.]com, a fake hiring page that encouraged users to schedule a call with an alleged FIFA recruiter. When a user tried to register with a personal Gmail address, the page asked for a business email instead. The page was steering victims toward credentials with more value: accounts that could provide access to a company environment.

<h4>The FIFA hiring page did not point to FIFA</h4>The fake page looked believable at first glance.

It used FIFA branding, presented a recruiter identity, and copied parts of the experience users see on legitimate recruiting and scheduling platforms. The recruiter’s name and photo appeared to match a real LinkedIn profile, which made the page feel more credible. For publication, we should keep that person’s identity redacted.The domain did not match FIFA’s real hiring infrastructure. FIFA’s legitimate careers pages are jobs.fifa.com and fifa.pinpointhq.com. The fake page used neither. It also loaded the favicon from the real FIFA recruitment site hosted at pinpointhq.com, giving the page a small visual cue borrowed from the legitimate service.Other signals pointed in the same direction. The domain and certificate were newly created, from May 23. The site used commodity hosting infrastructure, including Vercel and Render, rather than a corporate environment. Those signs are not enough on their own, but they fit the rest of the evidence.The login flow was the clearest part. Credentials were posted to:fifeq2026eqbackeq[.]onrender[.]com/api/loginThe Google sign-in prompt was not a real Google authentication window. It was rendered inside the page itself. The visible controls and links did nothing, and the page did not send the user into a genuine Google sign-in flow. According to the analysis, the only Google asset loaded was the favicon, apparently to make the fake window look familiar.A page can look like it opened a Google login without ever handing control to Google.Researchers also observed hxxps://careers-fifahiring[.]com, which later became inactive, and hxxps://fifajobs[.]com, which showed the same content and was indexed by Google.&nbsp;<h4>The recruitment lure went beyond FIFA</h4>The FIFA page was one version of a wider recruitment phishing pattern. The brand can change. The flow stays familiar: recruiter profile, meeting prompt, business email, fake sign-in.Researchers found similar fake hiring pages abusing other major brands, including Heineken, Hilton, Coca-Cola, Netflix, PepsiCo, Delta,&nbsp;and Spotify. In the Heineken case, the page appeared to use the identity of someone who, according to LinkedIn, really works as a recruiter at the company.

Additional hunting found two other recruitment-themed kits:hxxps://aquent-talent-jobs[.]com hxxps://hays-talent-opportunities[.]comThe Hays-themed version targeted Facebook credentials instead of Google credentials. Researchers assessed that the Aquent and Hays pages shared code, although they differed from the earlier FIFA examples.

This flexibility makes the campaign harder to frame as a single World Cup scam. FIFA provides the timely hook, but the mechanism is recruitment phishing. A scammer can swap the logo, recruiter name, and sign-in provider while keeping the same basic path to the victim’s credentials.<h4>Why fake recruitment pages work</h4>Recruitment gives attackers a clean excuse to ask for things that might otherwise raise suspicion.A candidate expects to click a scheduling link. They expect a recruiter profile. They expect a calendar prompt. In some cases, they may also expect to use a work email, especially for corporate, contractor, or partner-facing roles.The fake login window turns that ordinary flow into credential theft. Users are used to “Continue with Google” and “Continue with Facebook” buttons, but the button itself proves nothing. A genuine authentication flow should take the user to the provider’s domain and controls. A fake page can draw a convincing copy while keeping the victim inside the attacker-controlled site.The business email prompt is also a warning sign. A page that rejects a personal email and asks for a work address may be trying to raise the value of the stolen account.<h4>What job seekers should check before signing in</h4>Start from the company’s official careers page, not from a link in a message, ad, search result, or forwarded post. For FIFA, that means checking jobs.fifa.com or fifa.pinpointhq.com, not trusting a domain because it contains “fifa,” “career,” “hiring,” or “jobs.”Look closely at the domain before entering credentials. Scam pages often add plausible hiring words around a brand name: career, talent, jobs, hiring, portal, or connect. A polished page on a new domain is still a new domain.Treat the sign-in step as the real checkpoint. If a page offers Google or Facebook login, check whether the flow actually opens a real provider-controlled authentication page. A fake window embedded inside the original site, broken controls, inactive links, or a provider logo used only as decoration should stop the process.Use security software that includes web and phishing protection. These tools can block known malicious pages before they load, which is especially useful when scam domains look polished, appear in search results, or are shared through otherwise legitimate channels.<h4>What companies can do</h4>Recruitment pages are public by design, which makes them easy to copy.Security teams should monitor for newly registered domains combining their brand with hiring-related terms. They should also watch for pages that reuse favicons, recruiter photos, job descriptions, and calendar-booking language from legitimate hiring flows.HR and recruiting teams should make the official hiring path easy to verify. Candidates should not have to guess whether a domain is legitimate. A clear careers page, consistent recruiter communication, and warnings about unofficial scheduling links can remove some of the ambiguity attackers rely on.<h4>Conclusion</h4>The fake FIFA hiring page worked because it did not feel unusual. It looked like a normal step in a hiring process.Attackers do not need to invent new behavior when they can copy one people already trust. A branded scheduling page, a recruiter profile, and a familiar sign-in button can be enough to move a victim from interest to credential entry.The decision point is the login. Before entering a work email or password, check the domain, check where the authentication flow really goes, and verify the role through the company’s official careers site. Security software with web and phishing protection adds another layer, especially against short-lived scam pages that may only stay online for hours or days.IOCs:hxxps://fifahiring[.]com hxxps://careers-fifahiring[.]com hxxps://fifajobs[.]com hxxps://fifa-mg[.]shop hxxps://heineken-careers[.]com hxxps://aquent-talent-jobs[.]com hxxps://hays-talent-opportunities[.]com fifeq2026eqbackeq[.]onrender[.]com/api/login

Scammers are copying recruitment and calendar-booking flows to make fake Google and Facebook sign-ins look routine. 

Fake hiring pages abuse FIFA and other major brands to steal work credentials 

Threat Research Team Lead

Since January 2026, we have detected 7.4 million malicious incidents tied to residential proxy traffic, affecting 572,000 users in our telemetry. In this model, the household whose connection is being used is often not the buyer of the traffic. It is the exit point.That is the uncomfortable part of residential proxy networks. A home router, laptop, phone, smart TV, or other always-on device can quietly become part of a commercial network that routes strangers’ internet traffic through ordinary consumer IP addresses. To the websites on the other end, that traffic looks as if it came from a real household, because it did.The technology itself is not inherently malicious. Businesses use residential proxies for market research, ad verification, brand protection, security testing, and services that need to see the web as a real user would. But the same feature that makes these networks useful also makes them attractive for abuse: traffic from a household IP address is hard to block without risking false positives against legitimate users.The problem for consumers is simpler. Many people do not realize their device or connection can be enrolled into this kind of network at all. Some users opt in through pay-for-bandwidth apps. Others are enrolled through “free” VPNs, bundled software development kits inside unrelated apps, or unwanted software that installs proxy components without meaningful consent.Here is how the market works, who pays for it, and what it can cost the household that unknowingly hosts it.<h2>What is a residential proxy</h2>A residential proxy is an internet connection – typically a home laptop, phone, smart TV, or router – that has been enrolled into a third-party network and is being used to route someone else's traffic. To the rest of the internet, that traffic looks like it is coming from an ordinary household IP address assigned by a consumer ISP, not from a datacenter or&nbsp;VPN provider.Residential proxy networks are sold commercially. The model is two-sided: some people pay to tunnel their traffic out through someone else's connection (much like a VPN, except the exit point is a real home), while others get paid to provide those exit endpoints. Some are structured as paid peer-to-peer networks, where any participant's connection can serve as an exit for any other.Buyers pay for access to pools of millions of these "real user" IPs because they are extremely difficult to block, or rather, blocking them carries real consequences: the traffic originates from genuine households, so any block risks shutting out a legitimate user – effectively a false positive.Common buyers include market research firms, price-comparison services, brand-protection vendors, security teams testing their own defenses, and consumer privacy tools. At the abusive end of the market, the same infrastructure used by operators running automated purchasing bots, scrapping campaigns, ad fraud, credential stuffing, phishing and other attacks where hiding the true origin of the traffic matters. Demand is also being shaped by large-scale web scraping, including&nbsp;<a href="https://blog.includesecurity.com/2026/06/the-smart-tv-in-your-livingroom-is-a-node-in-the-aiscraping-economy/">scraping for AI training data</a>. As anti-bot systems become better at identifying traffic from datacenters and known VPN providers, residential IP addresses become more attractive because they look like ordinary users. That does not make every residential proxy user malicious. It does mean consumers may be carrying traffic for businesses and actors they have never heard of.<h2>How consumer devices end up in these networks</h2>There are several ways a home device can become part of a residential proxy network. Some are explicit. Others are easy to miss.The most obvious route is the pay-for-bandwidth model. Apps such as Honeygain, IPRoyal Pawns, and PacketStream pay users to share their internet connection. The user opts in, but they may not understand who will buy access to that connection later, what kind of traffic will pass through it, or what happens if that traffic triggers abuse reports.A second route is the “free” VPN or proxy app. Instead of charging a subscription, the app monetizes the user’s connection by reselling bandwidth. In practice, the user may think they are getting a privacy tool while also becoming infrastructure for other people’s traffic.A third route is through bundled software development kits, or SDKs. A developer adds a third-party component to an otherwise unrelated app, such as a flashlight, wallpaper, game, or utility. The developer gets paid when the app is active. The user sees the host app, not the proxy network behind it.The fourth route is malware or potentially unwanted software. In those cases, the proxy component is installed silently or deceptively, with no meaningful user consent at all.<h2>Why the "consent" is rarely meaningful</h2>Even when consent technically exists, it is often not the kind of consent most people would recognize.The disclosure may be buried in a long terms-of-service document. The wording may be vague, with phrases such as “share your unused bandwidth,” “support the app,” or “join our network.” These phrases do not clearly tell the user that arbitrary third parties may route traffic through the user’s home IP address.There is also the question of downstream use. Users are rarely told who buys access to the network or what those buyers do with it. Their connection could be used for benign market research. It could also be used for scraping, automated account creation, ad fraud, credential stuffing, or phishing infrastructure.The problem gets worse when the proxy component sits inside another app. The host app’s privacy policy may mention a partner SDK by brand name, but that still does not make it clear to a normal user that their device can become a proxy endpoint.Uninstalling the visible app may not always resolve the issue cleanly either, especially when proxy components are installed separately or bundled with other unwanted software.<h2>The risks consumers actually carry</h2>When a household device becomes a residential proxy node, the user takes on risks created by someone else’s traffic.The first risk is attribution. Abusive, fraudulent, or illegal activity routed through the device appears to come from the household’s IP address. ISP abuse notices, platform bans, fraud checks, and law enforcement requests point first to the subscriber, not to the person who bought the proxy traffic.The second risk is performance. Continuous third-party traffic can consume bandwidth, slow the home network, hit data caps, or violate ISP terms of service. The user may only notice that the internet feels slower, a bill is higher, or a service suddenly blocks access.The third risk is security exposure. Proxy clients run persistent background processes and maintain network connections so they can relay traffic. That adds always-on networking software to the device, often with limited visibility for the user.The fourth risk is reputation. A home IP address used for abuse can end up on blocklists used by banks, retailers, streaming services, and online platforms. The result can be account friction, failed logins, extra verification steps, or outright blocking from services the household actually uses.There is also a family angle. A child or another household member may install a free game, VPN, or utility without understanding the consequences. The legal, financial, and reputational risk still lands on the account holder.<h2>What our telemetry shows</h2>Since the beginning of 2026, we have detected 7.4 million malicious traffic incidents associated with residential proxy traffic, affecting 572,000 unique users in our user base.&nbsp; This is not a measurement of the entire residential proxy market. It reflects what we observe in our telemetry: malicious traffic, detections, and related activity affecting protected users. The real footprint is far larger, because residential proxy networks span major consumer platforms and many devices outside our visibility.The activity is global, but it is concentrated in a smaller set of countries. India leads by a wide margin, with roughly 83,000 affected users, followed by Vietnam with 61,000 and Brazil with 39,000. The Philippines (23,000), Poland (18,000), Indonesia (18,000) and Thailand (14,000) form a second tier, with Egypt, Spain, Ukraine, Argentina, Mexico and Italy each accounting for 10,000 to 13,000. The spread across South and Southeast Asia, Latin America, and Eastern and Southern Europe reflects where the host apps that carry this software are most widely installed.That geographic spread likely reflects a combination of app distribution, local install patterns, user appetite for free or pay-for-bandwidth apps, and the economics of bandwidth resale. It should not be read as a simple ranking of where residential proxy abuse originates.The chart below breaks down the threats we observed in this proxy-related traffic. Two categories dominate: phishing and malicious advertising. Behind them sits a long tail of consumer-facing fraud, including e-shop scams, generic scams, financial scams, and fake-tutorial pages, alongside outright malware such as trojans and other malicious traffic.&nbsp;

In the malicious proxy traffic we observed, scams and malware dominate. The important point for consumers is not only that their connection may be used by someone else. It is that, when abuse happens, their household IP address can be the return address.<h2>What consumers can do</h2>Most people will not identify a residential proxy component by name. But there are practical warning signs.Review “free” VPNs, proxy apps, bandwidth-sharing apps, and utilities from unknown publishers. Check whether they mention bandwidth sharing, peer-to-peer networking, traffic routing, or network participation in their terms. If an app offers payment for leaving your device online, assume your connection may be resold unless the provider explains exactly how traffic is vetted and who can buy access.On phones and computers, remove apps you do not recognize or no longer use. On home routers, check the list of connected devices and look for anything unfamiliar. If your ISP sends abuse notices, websites start blocking you, or your home IP address suddenly fails fraud checks, a proxy component is one possible explanation.Security software can also help by detecting unwanted proxy clients, bundled components, and related malicious activity. But this should not be a guessing game for consumers. Software that turns a household connection into a commercial exit node should be named clearly and made easy to remove.<h2>The takeaway&nbsp;</h2>With residential proxiesthe user’s connection becomes an infrastructure for someone else’s business.The issue is not that residential proxy technology exists. It has legitimate uses. The issue is whether people know when their devices are part of it, understand what kind of traffic may pass through their connection, and have a real choice to opt out.The fix is not to pretend consumers will read every EULA. This software should be named clearly, shown plainly, and made easy to remove before a household connection becomes someone else’s infrastructure.

Your home connection could be routing traffic for strangers. Here's how residential proxy networks work, how devices get enrolled and what our telemetry reveals about the risks for consumers.

Who's Really Using Your Home Internet Connection?

<h4>Key Points&nbsp;</h4><ul><li data-list-item-id="e43a143fd054561c50dba51071686adff">Gen Threat Labs has been tracking GoFlateLoader, a widespread Golang loader used to deliver multiple infostealers, including Amatera, Remus, Lumma, Vidar and StealC.&nbsp;</li><li data-list-item-id="e3d3b5d0814ea5b26aa75a8d3d6013a39">GoFlateLoader appears both in x86 (32-bit) and x86-64 (64-bit) variants, matching the bitness of the payload it is supposed to execute.&nbsp;</li><li data-list-item-id="e51ccafcfb8c418d894aeb685004a0ba5">The loader is designed for in-memory payload execution and is deliberately inflated with a massive PE overlay to hinder detection</li><li data-list-item-id="ec4bafef8a1113d24bc2b4b3a943228d1">The two main delivery paths are via supposedly cracked software and a <a href="https://research.checkpoint.com/2026/impersonation-click-hijacking-and-tds-inside-a-malware-distribution-ecosystem/" target="_blank" rel="noreferrer noopener">malicious TDS recently documented by Check Point Research</a>, which may redirect users to dedicated landing pages serving password-protected archives containing GoFlateLoader.&nbsp;</li></ul><h4>Introduction&nbsp;</h4>Not every threat that matters is technically sophisticated, and that is also the case with GoFlateLoader, which is a rather simple loader written in Go, whose sole purpose is to decode and execute the payload in memory. What stands out the most is not what the loader does but rather what it does not do – it comes without anti-debugging, anti-VM, or sandbox-evasion checks, and also lacks API hashing or CFG obfuscation, the kind of tricks that loaders almost always come with. Instead, GoFlateLoader relies on one of the simplest yet still effective tricks to stay under the radar – it appends a massive PE overlay at the end of the file, deliberately inflating the binary's size (hence the name GoFlateLoader).&nbsp;Nevertheless, it is important to understand that malware does not necessarily need to be elegant or stealthy. In the end, the only thing that matters to attackers is whether it can successfully deliver the final payload. And if that is the only metric, GoFlateLoader appears to be working fairly well for its operators, as we continue to see it widely distributed across our user base. Since the beginning of April, we have protected more than 33,000 unique users from it, with the most affected countries including Brazil, India, Argentina, Mexico, Turkey and Spain.&nbsp;

<h4>GoFlateLoader’s Execution Flow&nbsp;</h4>If we strip away the obfuscation noise that the Go compiler and the author’s decoy code add, the loader’s execution flow is short and linear:&nbsp;<ul type="disc"><li data-list-item-id="efa51af3b2859fd9c684efca0c82d3b66">Copy the encoded payload blob from the <code>.rdata</code> section onto the stack.&nbsp;</li><li data-list-item-id="e879bb970c6d1a81b8a42512222aceaa4">Decode the payload using a small, multi-stage, custom byte-level transformation into a valid PE.&nbsp;</li><li data-list-item-id="ebff4f340af15b75cbea0d682b26b7645">Parse the PE headers of the decoded payload (image base, image size, sections, data directories, etc.).&nbsp;</li><li data-list-item-id="e37fb26cad3d18a4a0c8bf9b2595d661f">Allocate an RWX memory region using <code>VirtualAlloc</code> with <code>MEM_COMMIT | MEM_RESERVE</code> and <code>PAGE_EXECUTE_READWRITE</code> flags, trying the payload’s preferred <code>ImageBase</code> first and falling back to an OS-assigned address if that fails.&nbsp;</li><li data-list-item-id="ebbde6d77cdc942c0a6bac79f4f8a6c28">Map the decoded payload into the allocated memory region by copying the PE headers first, then walking the section table and writing each section to its target virtual address.&nbsp;</li><li data-list-item-id="ed16738229e0f5c70a981c0c5f9b715f9">Apply relocations (only if preferred base mapping fails).&nbsp;</li><li data-list-item-id="e76cde4bae1be6c21b897c350c6fd15db">Resolve imports by walking the standard <code>IMAGE_IMPORT_DESCRIPTOR</code> table and rebuild the IAT in place via <code>LoadLibrary</code> and<code> GetProcAddress</code>.&nbsp;</li><li data-list-item-id="e786887a3d34fc7ce344ee8375944799e">Transfer execution to the final payload using Go’s <code>syscall.Syscall</code>.&nbsp;</li></ul>&nbsp;This is essentially a textbook example of manual PE loading with the payload reconstructed and executed entirely in memory so that it never touches the disk.&nbsp;&nbsp;

Notably, GoFlateLoader transfers execution to the final payload via <code>syscall.Syscall</code>. In Go, <code>syscall.Syscall</code> has the signature <code>Syscall(trap, nargs, a1, a2, a3)</code>, where <code>trap</code> is the function pointer to invoke, <code>nargs </code>is the argument count, and <code>a1</code>, <code>a2</code>, and <code>a3</code> are the first three arguments passed to the target function. Here, trap is set to point to the final payload’s entry point, <code>nargs</code> is set to <code>1</code>, and since the execution is transferred to that entry point, the remaining arguments serve only as filler values and are never used by the invoked payload. In other words, GoFlateLoader is not issuing a <code>syscall</code> at all, but abusing <code>syscall.Syscall</code> as a generic call gate to hand control to the payload.&nbsp;&nbsp;It is worth emphasizing that GoFlateLoader consistently sets these arguments to the hardcoded values <code>1</code>, <code>2</code>, <code>3</code>, and <code>4</code>, which is unusual enough to stand out as a potential detection opportunity (combined with other patterns).&nbsp;

The loader also carries substantial junk/decoy code, which varies across builds and serves no purpose other than padding the control flow and complicating static analysis.&nbsp;&nbsp;&nbsp;<h4>Inflated PE Overlay&nbsp;</h4>GoFlateLoader’s defining characteristic is its massive size (typically 700-950 MB), caused by an artificially inflated PE overlay appended to the original loader.&nbsp;

This is a deliberate design choice intended to bypass detection, as excessively large executable files may pose challenges for AVs and EDRs, primarily due to performance and resource constraints. As a result, to preserve performance and avoid noticeable impact to end users, these solutions often enforce practical file-size limits for deep scanning and/or emulation.&nbsp;The same logic extends to automated analysis pipelines – bloated files frequently cause timeouts during heuristic analysis and effectively prevent the sample from being uploaded to cloud-based sandboxes or threat intelligence platforms, which enforce their own strict upload limits to manage bandwidth and storage costs. A prime example is VirusTotal, which enforces a strict 650 MB upload limit. The fact that GoFlateLoader consistently sits just above this threshold strongly suggests that VirusTotal is one of the platforms it specifically aims to evade.&nbsp;&nbsp;In most cases, GoFlateLoader’s PE overlays are filled with null bytes, though we have also observed builds with random padding.&nbsp;&nbsp;

Crucially, adding such a PE overlay comes at virtually no distribution cost for attackers, as the inflated data compresses to a fraction of its original size (assuming constant padding).&nbsp;&nbsp;<h4>Delivered Final Payloads&nbsp;</h4>The final payloads distributed through GoFlateLoader are predominantly information stealers, with the most prevalent currently being Amatera, <a href="https://www.gendigital.com/blog/insights/research/remus-64bit-variant-of-lumma-stealer" target="_blank" rel="noreferrer noopener">Remus</a>, and Lumma. Beyond these, we have also observed it distributing Vidar, StealC, and SvitStealer.&nbsp;GoFlateLoader comes in both x86 and x86-64 variants, with each variant matched to the architecture of the payload it is intended to execute.&nbsp;&nbsp;As for how GoFlateLoader binaries are distributed, we have observed at least two notable delivery paths. The first is through supposedly cracked software. The second involves the payload being served via <a href="https://research.checkpoint.com/2026/impersonation-click-hijacking-and-tds-inside-a-malware-distribution-ecosystem/" target="_blank" rel="noreferrer noopener">the malicious TDS recently described by Check Point Research</a>, which redirects users to a landing page that offers a downloadable archive and separately displays the password required to extract it. This approach provides several advantages for the threat actor. First, compression significantly reduces the binary size due to the constant-padded PE overlay. Second, because the archive password is displayed separately on the landing page, it makes it harder for AV and EDR solutions to obtain it and decrypt the archive. As a result, the payload may remain concealed until the victim manually extracts it.&nbsp;&nbsp;&nbsp;<h4>Summary&nbsp;</h4>GoFlateLoader is a reminder that prevalence and sophistication are not the same thing. At its core, it is a rather simple manual in-memory PE loader written in Go, with no novel anti-analysis tricks under the hood. Whatever evasion it achieves comes not from obfuscation or other anti-analysis tricks but from its size – the appended PE overlay is built to slip past size-constrained scanners and automated analysis pipelines. Yet, despite its low level of sophistication, GoFlateLoader remains an active and relevant threat, used to deliver a wide range of prevalent information stealers.&nbsp;&nbsp;Gen products protect our users against both GoFlateLoader and the payloads it delivers.&nbsp;&nbsp;<h4>Indicators of Compromise (IoCs)&nbsp;</h4>Note: The files listed in the IoCs that are not archives all exceed 650 MB in size and are therefore not available on VirusTotal.&nbsp;b88c5744975d2abb447aecc6c090fee9f8580413f4612eecdc6ed1973e8a1739 (password-protected archive containing GoFlateLoader x64 variant loading Remus; pwd: 1234)&nbsp;ed5ae7f36453c5a23e9868a5729d67e0549a11f6dea54f5f52d654a8f51d4902 (archive containing GoFlateLoader x64 variant loading Remus)&nbsp;841c9297cb8a2e0ff89433d13c05bfc760eb2e98e251cb8fa785d2ad7cbac05f (archive containing GoFlateLoader x86 variant loading Amatera)&nbsp;ece7c48eb411b24f26762ede83badb4a644c41d5777129381ac2541804d64fc2 (archive containing GoFlateLoader x86 variant loading Lumma)&nbsp;421ce2d2f49c23bbe9f60ef3b9cd38d7eb912ce02e56a61837656210069bd9e2 (archive containing GoFlateLoader x64 variant loading Vidar)&nbsp;121c2dc793b3873f75a29ec02241f94136de19c049382a50a50d0d5b99507073 (GoFlateLoader x64 variant loading StealC)&nbsp;2415db5081cec9bfd14ad6da1a66169fd96f13a49010c319a73d1ed6fafd4efa (GoFlateLoader x64 variant loading Vidar)&nbsp;d9917ade3b4c125a95b5d3e6343cde26145dfbf569bd7e2a843fd0c6fc8ddc28 (GoFlateLoader x64 variant loading Remus)&nbsp;4cf6893756f441522b94b36f10e5de0e47aeed4743f95c51650746d1ecf97e3d (GoFlateLoader x64 variant loading SvitStealer)&nbsp;8b89d6c9152d3aab97aadd515ecb69ca72654db2f25425759ba4b646853d737d (GoFlateLoader x86 variant loading Lumma)&nbsp;90ce4ff9da23ac150da0a8e17930cab1e369aa349fdc1b65691b70369145664a (GoFlateLoader x86 variant loading Amatera)&nbsp;

 Intentionally Inflated Go Binaries That Use Massive PE Overlays to Bypass Detection

GoFlateLoader: A Widespread Golang Loader Delivering Multiple Infostealers

<h2 class="text-align-justify">Key Points</h2><ul><li>Gen Threat Labs has discovered and analyzed an elaborate Traffic Direction System infrastructure delivering FakeCaptcha, tech scams and other malware campaigns to select users.&nbsp;</li><li>The campaign entry points are infected or otherwise attacker-controlled streaming websites, file sharing services, as well as malvertising campaigns.&nbsp;</li><li>Victims are evaluated based on geolocation, IP address, and browser fingerprinting; for example, connections through VPNs or headless browsers are detected and rejected.&nbsp;</li><li>The FakeCaptcha campaign is increasing its stealth by mimicking legitimate software websites.&nbsp;</li><li>A new variant of FakeCaptcha spreading information stealers and RATs avoids detection by employing Unicode math fonts.&nbsp;</li></ul>

<h2>Introduction</h2>The <a href="https://www.gendigital.com/blog/insights/research/global-surge-in-fake-captcha-attacks" target="_blank" rel="noreferrer noopener">FakeCaptcha campaigns</a> have become a widespread malware infection vector, using clever social engineering and exploiting internet users’ confusion about common CAPTCHA practices to trick victims into infecting their own computers, typically with information-stealing malware like LummaC2, by pasting malicious commands into the Windows Run dialog.&nbsp;We have discovered an elaborate infrastructure actively used to deliver multiple variants of FakeCaptcha and other malicious content to select users. At the core of the operation is an attacker-controlled Traffic Direction System network. It has a single purpose—to fingerprint the visitor’s device and network details to determine what type of content to deliver. This could be malicious, monetized or decoy content, or no content at all. For reasons we will describe later, we are calling this infrastructure HelloTDS.

As shown in the graphic above, the scope of the campaign is massive; in April and May 2025 alone, we have protected over 4.3 million users worldwide. In terms of absolute numbers, the most affected regions include the United States, Brazil, India, and Western Europe. However, when adjusting for population size and user base, a different trend emerges: the highest relative risk is concentrated in the Balkan countries of Europe and in parts of Africa, particularly Rwanda, Egypt, Tanzania, and Kenya.<h2>Attack Chain</h2>At a high level, the attack works as follows: The victim visits an infected website or clicks on a malicious ad. The entry website requests a JavaScript file from the initial HelloTDS endpoint, which fingerprints the connection and decides whether to serve content or stay idle. If the user is not deemed a suitable victim, the attack stops here. Otherwise, the server responds with a script that further fingerprints the browser, logs mouse movements and similar, eventually sending another request to receive the final payload containing a redirect to the malicious landing page. Among users who passed the initial check, FakeCaptcha was observed to be the most prevalent malicious landing page, with Fake Updates, file downloads containing encrypted malware, tech scams and other fraudulent campaigns also having a share.&nbsp;

One specific example of the full attack chain is illustrated in the next figure: We visited the website dailyuploads[.]net, uploaded an arbitrary image file, and navigated to the link generated by the service. The resulting page included a script from HelloTDS, and when we clicked the download link, we were involuntarily redirected to a FakeCaptcha landing page attempting to trick visitors into executing the Emmenhtal loader.

<h2>From Movie Streaming or File Sharing to Malware</h2>Besides ad networks, the primary source of traffic for HelloTDS are legitimate-looking, properly functioning websites, such as movie streaming sites, torrent mirrors, file-sharing services, link shorteners, or porn websites. However, these websites are specifically crafted or manipulated—presumably also by the attacker—to cooperate with the attacker’s infrastructure and load malicious scripts. This tactic completely bypasses the need for malvertising campaigns through “legitimate” advertising companies, as these websites request the attacker’s scripts directly.Moreover, we have observed a repeating pattern in some of these websites, which offer file sharing services and promise their users financial gains. These include dailyuploads[.]net, streamtape[.]{com,net,to}, watchadsontape[.]com, bigwarp[.]art or savefiles[.]com to name a few. Security or data privacy is clearly not a priority for these services, as anyone can sign up with any e-mail address without verification. We suspect that these websites are cloned from some existing file sharing service and controlled by the same threat actor who controls HelloTDS and eventually delivers FakeCaptcha, for the primary purpose of embedding malicious popups and redirects within the page (although email &amp; password harvesting could be a “welcome” byproduct).

<h2>The Anatomy of a HelloTDS Domain</h2>An “entry point” website, like those described previously, will usually statically embed one or more HTML snippets loading remote JavaScript that look like this:

(In case of file-sharing websites described above, this snippet will be included in the page with the uploaded file, i.e. the one you share the link to, but not on the main page itself.)The source domain is randomly rotated and both path segments are randomly generated (although their length and character sets are always the same), presumably to prevent static detection. Despite this, it is easy to spot these URLs with a human eye (or a clustering algorithm). All the HelloTDS domains we have seen so far have the same format: The apex domain consists of 2-3 (pseudo)words with a “.top”, “.shop” or sometimes “.com” TLD; optionally, a short subdomain is used, consisting of usually 2, but up to 4 random letters.&nbsp;Examples of these domains include:<ul><li>yr[.]unasonoric[.]com</li><li>gq[.]binesyorker[.]com</li><li>sb[.]rowlandpodogyn[.]shop</li><li>nutatedtriol[.]com</li></ul>

Apart from the distinctive domain names, another sign that speaks for HelloTDS is that the domain names are almost exclusively registered in Panama through the Pananames registrar, and the servers reliably present Let’s Encrypt certificates during TLS handshakes.In an attempt to make tracking HelloTDS infrastructure difficult, the authors designed it to be highly dynamic, frequently rotating active domains and their DNS records. Even so, we have identified that all domains associated with this infrastructure are registered under AS7979 (SERVERS-COM) within the following IP ranges:<ul><li>23.83.64.0/22</li><li>23.109.120.0/22</li><li>23.109.128.0/19</li><li>23.109.170.0/24</li><li>94.242.232.0/21</li><li>172.240.104.0/21</li><li>172.241.48.0/21</li><li>172.255.96.0/20</li><li>173.0.146.0/23</li><li>173.237.64.0/20</li><li>188.42.104.0/21</li><li>188.42.240.0/21</li><li>209.192.192.0/18</li><li>212.117.184.0/21&nbsp;</li></ul>

Another interesting aspect shared by all HelloTDS servers is the usage of the non-standard HTTP header “megageocheckolololo”, which we believe is not commonly used by any well-known service. Its presence in the “Access-Control-Allow-Headers” HTTP response header is another indicator of a HelloTDS server.

Lastly, domains belonging to HelloTDS can be reliably identified using the information they expose via “robots.txt”:

The “/hi” endpoint always returns a simple HTML response with a single text node that says “Hello!” — hence the name HelloTDS.<h2> First Stage: Fingerprinting Based on Network Information</h2>The first decision is made by the HelloTDS server immediately upon receiving the request sent from one of the entry point websites. Since we can only send requests and observe the servers’ responses, it is difficult to tell the exact shape of the server’s decision tree. We suspect that several factors, such as IP/geolocation checks, timeouts, or maybe even pure randomness, may contribute to the result in some way.From the user’s point of view, there are two possible outcomes. In one case, the server responds with a no-op script (usually an empty function or a single-line comment containing 2 or 3 letters, like “// ipx” or “// mt”, likely codes used for “debugging” the TDS), meaning the user will not get any malicious content (from that endpoint). Alternatively, the server will respond with a large JavaScript that performs heavy client-side fingerprinting and eventually performs a subsequent request to another HelloTDS endpoint. Either way, the server sets a pair of cookies in the response; these encode information about the client’s browser, operating system and network, as detected by the server, and may be used as a client-side cache for further requests (although we have found that tampering with them is generally not useful).

The cookie values contain Zlib-compressed binary data encoded in Base64. Although the low-level data format is unknown, the data itself contains readable ASCII strings like the client’s IP address, country, or ISP, which reveal the purpose these cookies likely serve.

<h2>Second Stage: Fingerprinting in the Browser</h2>Contrary to the logic executed on the server, the next fingerprinting routine is performed in the browser (by the returned JavaScript) and can thus be reverse-engineered. The exact code returned from different HelloTDS endpoints varies somewhat in structure and complexity (at the very least, the server appends a timestamp comment to hinder hash-based detection), but all of them share the same basic high-level functionality.Crucially, the JavaScript sent by the server contains an encrypted configuration variable — a JSON string where each alphanumeric character is substituted with some other. This part of the code can be easily identified: Look for a function call with two arguments, the first being the encrypted JSON string and the second being the key: A concatenation of the characters a-z, 0-9 and a random permutation thereof. This obfuscation is not unique to HelloTDS; however suspicious it may be for an ad network to employ this technique, we have also observed this behaviour in scripts employed by the Czechia-based Clickadu ad network, which happens to share certain characteristics with the threat actor’s infrastructure.The number of key-value pairs in the configuration object varies, but some common ones we regularly encounter are “url”, “zone_id”, “uuid_url”, “rot_url”, “metric_url”, “timezone_offset” and/or “timezone_diff”. Other keys that particularly stand out due to their names are “devtools_protection” and “philanthropic_level”, which were set to “true” and 0 respectively whenever we inspected a chain that landed on FakeCaptcha in the end.The “*_url” keys contain further HelloTDS URLs, which are used to retrieve a sort of unique identifier (“uuid_url”), the target URL (“rot_url”), an endpoint to report errors and other metrics to (“metric_url”), and an endpoint for mouse movement and device sensors tracking (“url”). Notably, the “rot_url” always contains a special parameter named “md” with a placeholder for the fingerprinting information collected by the script.

The JavaScript collects the following information:<ul><li>Basic browser info: window width and height, document referrer, window location (URL), browser language, current datetime, time zone offset, and whether the site is shown in a top-level window or an iframe,</li><li>the “user id” previously returned by the UUID endpoint,</li><li>whether or not a mobile browser is used (by checking if the deprecated window.orientation property is defined),</li><li>whether or not a touch screen is present (by reporting the value of the window.navigator.maxTouchPoints property),</li><li>page title information, most common words in the host webpage as well as any keywords declared in the host document’s head element,</li><li>the approximate amount of physical memory and the number of processor cores,</li><li>battery status (level, charging state),</li><li>WebGL vendor and renderer,</li><li>a “sandbox score” consisting of 6 flags, e.g., whether the browser is Headless Chrome, or whether specific video codecs are supported,</li><li>network information, e.g., network type (Wi-Fi, Ethernet, cellular, …), effective and maximum bandwidth, or the estimated round-trip time,</li><li>screen orientation and colour depth,</li><li>whether cookies are enabled.</li></ul>

A JSON object containing this information (as well as some keys from the config) is Base64-encoded and sent in the md parameter to the rotated URL from the config via a POST request. The response contains a JSON object with a “url” key as well as a TTL and bidding details.

The returned URL is the last HelloTDS endpoint in the whole chain.<h2>Third Stage: Final Payload with Redirect to Landing Page</h2>The final step in the HelloTDS pipeline is a simple JavaScript-based redirect to the target URL.

In this case, bestfree4u[.]com simply responded with a 301 redirect to a FakeCaptcha landing page at finding-from-internet[.]fly[.]storage[.]tigris[.]dev/seacrh-result[.]html. However, when accessed from a different environment, the same domain may seem harmless. In other cases, the final HelloTDS redirect leads to tech support scams, malicious file downloads, fake browser updates, potentially unwanted browser extensions, or cryptocurrency investment platforms.<h2>The Destination: FakeCaptcha Redirectors</h2>In the past months, we have seen a highly predictable pattern in the malicious URL payloads served by HelloTDS: Almost universally, they were machine-generated domains consisting of a couple of English words and a “.com”, “.shop”, “.online”, “.info”, “.live”, “.pro” or similar TLD, and the URL path was a string with 24 hexadecimal digits, the first two usually being 67 or 68. These URLs served as simple redirects, either to FakeCaptcha, or sometimes to a decoy—usually some (benign but potentially unwanted) browser extension or cryptocurrency product website. We have seen multiple of these URLs redirect to both a FakeCaptcha landing page and a decoy at different times, which removes any doubt about this being an inherently malicious service. Example URLs include the following:<ul><li>actednow[.]com/675cb495c39bc481ddf8edd6,</li><li>buzzflying[.]shop/6767af2ee2aa535e92d62a64,</li><li>goldtera[.]live/6758b1d6467532a801fa06c4,</li><li>orbito[.]online/677120fb2cca41d88a7392a2.</li></ul>The domains used by this service could be reliably identified by being registered under Namecheap in Reykjavik, Iceland, using a Let’s Encrypt TLS certificate, and resolving to one of the following IP addresses:<ul><li>5.161.37.228,</li><li>5.161.110.119,</li><li>5.161.178.111,</li><li>23.105.163.27,</li><li>23.105.163.55,</li><li>34.142.174.238,</li><li>172.104.80.249.</li></ul>Recently, the attacker has started employing more sophisticated masking even at the level of the final payload.Additionally, the structure of served URLs redirecting to FakeCaptcha has changed – they now often include parameters with placeholders, which may not even be filled in, yet the victim can still be redirected to a FakeCaptcha landing page. We have seen different variations of these URLs, some of which omit specific parameters.

Interestingly, some of the URLs served by HelloTDS redirect victims to FakeCaptcha landing pages, but when accessed in a controlled environment, they often mimic some legitimate websites and serve some benign content. One example of such a website is&nbsp;hxxps[://]avs4u[.]net/?sub1=39388&amp;sub2=…&amp;sub3=ID&amp;sub4=677886&amp;sub5=0.0004752 — while it often redirects the victims to FakeCaptcha landing pages, it may also display mimicked content of the legitimate website www.avs4you.com, a software for processing videos and images. The downloadable binary is also legitimate, indicating an attempt to conceal malicious activity.Another example of a URL redirecting victims to FakeCaptcha but serving legitimate content in a controlled environment is hxxps[://]arcadeclassic[.]org/?sub1=681978&amp;sub2=0.00037565.&nbsp;

We therefore tried to understand the reasoning behind the decision of whether malicious content is served. As part of this effort, we explored the potential influence of the cookies being set, which we mentioned earlier. Ultimately, we have concluded that the decision is probably based solely on the IP address of the client. We suspect that if the IP address originates from a known VPN or other anonymizing services, the victim is consistently redirected to some benign content, likely because such users are not target victims and may potentially be security analysts. In such cases, HelloTDS delivers benign content to obscure its actual behavior, effectively evading analysis and avoiding detection.<h2>Delivered Malicious Content</h2>In our analysis of HelloTDS landing page redirectors, we observed that they serve a mix of both malicious and legitimate content. These “clean” redirects, however, often lead to investment-related sites, especially those involving cryptocurrencies. We believe that this choice is deliberate. Firstly, generating traffic to advertisers through non-consensual clickjacking may serve as an alternative, more predictable source of income for the attacker without “stepping into the dark” and serving hard malware. Furthermore, if the primary objective of FakeCaptcha is to infect victims with information stealers that ubiquitously target crypto wallets, redirecting users who were not deemed as ideal victims to a crypto investment platform may be the most logical strategy for the attacker in the long run.On the unequivocally malicious side, various forms of FakeCaptcha were predominant, but we have also encountered Fake Updates, tech scams, download links with encrypted archives delivering LummaC2, and various other scams. Some specific examples are listed below.

<h2>Key Takeaways and Defensive Measures</h2>The HelloTDS infrastructure behind FakeCaptcha campaigns demonstrates how attackers continue to refine their methods to bypass traditional protections, evade detection, and selectively target victims. By leveraging sophisticated fingerprinting, dynamic domain infrastructure, and deception tactics (such as mimicking legitimate websites and serving benign content to researchers) these campaigns achieve both stealth and scale.What You Can Do:<ul><li>Use reputable security software with real-time protection to block known TDS and FakeCaptcha domains. Products from Norton, Avast, AVG, and Avira automatically protect users against these threats.</li><li>Enable browser-level protections like anti-tracking and ad-blocking extensions, which may prevent the initial malicious scripts from loading.</li><li>Be cautious with file-sharing and streaming sites, especially those that promise monetary rewards or allow anonymous uploads without verification.</li><li>Avoid copying and pasting commands from unknown or suspicious websites into system dialogs like Windows Run.</li><li>Inspect suspicious domains manually or through security tools—look for signs like oddly-structured URLs, frequent domain rotation, or generic domain names.</li></ul>FakeCaptcha is not just a social engineering trick. It's the final step in a well-orchestrated system that filters, evaluates, and targets users with precision. Understanding the infrastructure—HelloTDS in this case—is key to disrupting the full chain of attack.<h2>Indicators of Compromise (non-exhaustive)</h2>You can find the complete list of IoCs on our <a href="https://github.com/avast/ioc/tree/master/HelloTDS">GitHub</a>.<h3>File Sharing Entry Websites</h3>dailyuploads[.]netstreamtape[.]tostreamtape[.]netstreamtape[.]towatchadsontape[.]combigwarp[.]artsavefiles[.]com<h3>HelloTDS Domains</h3>yr[.]unasonoric[.]comgq[.]binesyorker[.]comsb[.]rowlandpodogyn[.]shopnutatedtriol[.]commixscoggan[.]shopbu[.]unrimedironize[.]shop<h3>FakeCaptcha Redirectors</h3>actednow[.]com/675cb495c39bc481ddf8edd6buzzflying[.]shop/6767af2ee2aa535e92d62a64goldtera[.]live/6758b1d6467532a801fa06c4orbito[.]online/677120fb2cca41d88a7392a2bestfree4u[.]comavs4u[.]net/arcadeclassic[.]org<h3>FakeCaptcha Landing Pages</h3>adelaidavizcaino[.]com/cpwpartage-de-medias[.]fly[.]storage[.]tigris[.]dev/affiliate-link[.]htmlfinding-from-internet[.]fly[.]storage[.]tigris[.]dev/seacrh-result[.]html

A behind-the-scenes look at HelloTDS: the stealthy infrastructure driving FakeCaptcha scams, malware and more across millions of devices

Say hi to HelloTDS: The infrastructure behind FakeCaptcha 

Threat Analysis Engineer at Gen

Imagine receiving a message on LinkedIn from a recruiter offering an enticing opportunity: a well-paid front-end development project that seems almost too good to be true. It’s a scenario many aspiring programmers dream of, especially those eager to land their first job in the tech industry. However, as this story reveals, the dream can quickly turn into a cybersecurity nightmare.&nbsp;<h2>The Setup&nbsp;</h2>The attack begins innocuously enough. A recruiter reaches out on LinkedIn, pitching an appealing development project with an impressive salary. The target? Entry-level front-end developers, often beginners in the field and eager to seize opportunities. Once the victim expresses interest, the recruiter provides a link to a code repository, ostensibly containing the project files.&nbsp;In one known case, the recruiter’s high salary offer raised suspicions. This wariness led the targeted individual to scrutinize the repository, revealing malware embedded within the code. The victim wisely decided against proceeding, avoiding infection entirely. Shortly thereafter, the recruiter’s LinkedIn profile disappeared, leaving questions about their next steps unanswered.&nbsp;<h2>A Firsthand Example&nbsp;</h2>This story came to light when we became aware of a similar job offer on LinkedIn. The recruiter claimed to be developing a decentralized application (DApp) for purchasing and claiming ownership of video games using NFTs. The project aimed to provide a seamless cross-platform experience and was allegedly seeking skilled web developers to help with their UI/UX redesign.&nbsp;

After a few messages, the recruiter shared a Bitbucket repository link. At first glance, the project appeared legitimate, with standard documentation and architecture diagrams detailing the NFT-based system. However, deeper investigation revealed a suspicious obfuscated JavaScript file, onfig-overrides.js, containing malicious code. This file, overlooked by an untrained eye, was designed to:&nbsp;<ul><li>Steal login credentials</li><li>Exfiltrate crypto wallet data</li><li>Deploy and execute a Python loader&nbsp;</li></ul>The recruiter’s polished approach and seemingly genuine project made the scam particularly deceptive.&nbsp;

<h2>The Mechanism&nbsp;</h2>The attack exploits standard industry practices. In many coding interviews or preparation tasks, candidates are asked to clone a repository, run a local server, and work on the application offline. This is commonplace for front-end developers testing web applications. However, in this case, the repository included concealed malware designed to compromise the developer’s system.&nbsp;<h2>The Step-by-Step Anatomy of the Attack&nbsp;</h2>Once executed, the obfuscated JavaScript file initiated a multi-stage attack process, consisting of two main phases: the initial JavaScript execution and the subsequent Python-based payloads.&nbsp;Phase 1: JavaScript Execution&nbsp;<ol><li>Information Stealing: The initial script acted as an infostealer, targeting:</li></ol><ul><li>Solana wallet configurations and stored crypto wallet data.</li><li>Login credentials stored in web browsers such as Firefox and Chrome.</li><li>Other sensitive information that could be used for further attacks.</li></ul><ol start="2"><li>Python Loader Deployment: The JavaScript file downloaded and executed a Python-based loader, which was responsible for delivering the next set of malicious payloads.&nbsp;</li></ol>Phase 2: Python-Based Payloads&nbsp;<ol start="3"><li>Remote Access Trojan (RAT): The Python loader executed a RAT with keylogging and advanced spying capabilities, including:</li></ol><ul><li>Remote shell commands: Enabling attackers to execute commands on the victim’s machine.</li><li>Clipboard exfiltration: Capturing and transmitting clipboard data to the attacker.</li><li>Payload downloads: Downloading and executing additional malicious files.</li><li>Browser process termination: Killing Chrome and Brave browser processes to disrupt user activity.</li><li>Sensitive file exfiltration: Searching for and exfiltrating critical files such as .env, readme, and .git.</li></ul><ol start="4"><li>XMRig Cryptominer: As part of the second Python stage, the XMRig cryptominer was deployed, exploiting the victim’s computing resources to mine cryptocurrency, which resulted in degraded system performance and increased power consumption.&nbsp;</li></ol>This structured attack exemplifies the multi-layered and highly intrusive nature of this attack, ilustrating how it moves beyond simple credential theft to fully compromising the victim’s machine.&nbsp;<h2>The Lazarus Group Connection?&nbsp;</h2>Interestingly, the tools used in this attack, named "BeaverTail" and "InvisibleFerret," have been attributed to the notorious Lazarus group in the past. We have strong indications that the attack was orchestrated by the Lazarus group, given their history of sophisticated cyber campaigns and their association with the tools used. The attackers leveraged tools previously linked to high-profile cybercrime, making it critical to approach such situations with caution. While the involvement of Lazarus cannot be confirmed with absolute certainty, the evidence strongly suggests their involvement.&nbsp;<h2>The Broader Implications&nbsp;</h2>While this specific attack focused on credential theft and cryptojacking, the Remote Access Trojan (RAT) also allows attackers to maintain long-term control over infected systems. This could be used for espionage, further malware deployment, or data exfiltration. Additionally, the potential harm could extend far beyond. The perpetrators could use such schemes to:&nbsp;<ul><li>Compromise company systems through stolen credentials, allowing further infiltration and unauthorized access.</li><li>Utilize the compromised system for ransomware deployment or additional malware infections.</li><li>Leverage the infected machine as a launching point for future cyberattacks, amplifying the scope of the breach.&nbsp;</li></ul>The idea of attackers genuinely hiring developers to create malicious tools is not far-fetched, though it seems unlikely in this instance.&nbsp;<h2>Lessons for Job Seekers&nbsp;</h2>This incident serves as a crucial reminder for job seekers, particularly those new to the field, to exercise vigilance when assessing job offers. Here are some tips to protect yourself:&nbsp;<ol><li>Have an online interview before proceeding: Before starting any test or downloading files, request a video call with the recruiter to verify their legitimacy and discuss the job offer in detail.</li><li>Verify the recruiter’s credentials: Cross-check their profile and company information. If something seems off, trust your instincts.</li><li>Scrutinize repositories: Before cloning or running any code, scan the files for anomalies. Use reputable security tools to analyze suspicious files.</li><li>Question too-good-to-be-true offers: Unrealistically high salaries or overly eager recruiters can be red flags.</li><li>Use sandbox environments: Run unfamiliar code in isolated environments to prevent potential infections.</li><li>Stay informed: Keep up with the latest cybersecurity threats to recognize warning signs early.&nbsp;</li></ol><h2>Conclusion&nbsp;</h2>Threat actors are increasingly targeting individuals with highly personalized and sophisticated methods, exploiting both technical vulnerabilities and human psychology. This case serves as a sobering reminder of the lengths cybercriminals will go to exploit human trust and technological processes. For aspiring developers and job seekers, awareness and skepticism are critical defenses. As we face increasingly sophisticated threats, a collective effort to educate and inform will play a crucial role in minimizing risks. By staying vigilant, leveraging security tools, and promoting best practices, we can create a safer digital landscape for everyone—one where opportunities are genuine, and trust is protected.&nbsp;

Milan Špinka

Posts by Milan Špinka

Say hi to HelloTDS: The infrastructure behind FakeCaptcha

How Cybercriminals Are Turning Job Offers Into Malware Traps