Cybersecurity researchers have uncovered a widespread phishing marketing campaign that makes use of faux CAPTCHA photographs shared through PDF paperwork hosted on Webflow’s content material supply community (CDN) to ship the Lumma stealer malware.
Netskope Risk Labs stated it found 260 distinctive domains internet hosting 5,000 phishing PDF information that redirect victims to malicious web sites.
“The attacker makes use of search engine optimization to trick victims into visiting the pages by clicking on malicious search engine outcomes,” safety researcher Jan Michael Alcantara stated in a report shared with The Hacker Information.
“Whereas most phishing pages give attention to stealing bank card info, some PDF information include faux CAPTCHAs that trick victims into executing malicious PowerShell instructions, in the end resulting in the Lumma Stealer malware.”
The phishing marketing campaign is estimated to have affected greater than 1,150 organizations and greater than 7,000 customers for the reason that second half of 2024, with the assaults primarily singling out victims in North America, Asia, and Southern Europe throughout expertise, monetary providers, and manufacturing sectors.
Of the 260 domains recognized to host the faux PDFs, a majority of them are associated to Webflow, adopted by these associated to GoDaddy, Strikingly, Wix, and Fastly.
Attackers have additionally been noticed importing a number of the PDF information to legit on-line libraries and PDF repositories like PDFCOFFEE, PDF4PRO, PDFBean, and Web Archive, such that customers looking for PDF paperwork on serps are directed to them.
The PDFs include fraudulent CAPTCHA photographs that act as a conduit to steal bank card info. Alternatively, these distributing Lumma Stealer include photographs to obtain the doc that, when clicked, takes the sufferer to a malicious web site.
For its half, the location masquerades as a faux CAPTCHA verification web page that employs the ClickFix method to deceive the sufferer into operating an MSHTA command that executes the stealer malware by way of a PowerShell script.
In latest weeks, Lumma Stealer has additionally been disguised as Roblox video games and a cracked model of the Whole Commander instrument for Home windows, highlighting the myriad supply mechanisms adopted by numerous risk actors. Customers are redirected to those web sites by means of YouTube movies possible uploaded from beforehand compromised accounts.
“Malicious hyperlinks and contaminated information are sometimes disguised in [YouTube videos, comments, or descriptions,” Silent Push said. “Exercising caution and being skeptical of unverified sources when interacting with YouTube content, especially when prompted to download or click on links, can help protect against these growing threats.”
The cybersecurity company further found that Lumma Stealer logs are being shared for free on a relatively new hacking forum called Leaky[.]professional that went operational in late December 2024.
Lumma Stealer is a fully-featured crimeware answer that is provided on the market underneath the malware-as-a-service (MaaS) mannequin, giving a approach to harvest a variety of data from compromised Home windows hosts. In early 2024, the malware operators introduced an integration with a Golang-based proxy malware named GhostSocks.
“The addition of a SOCKS5 backconnect characteristic to current Lumma infections, or any malware for that matter, is very profitable for risk actors,” Infrawatch stated.
“By leveraging victims’ web connections, attackers can bypass geographic restrictions and IP-based integrity checks, significantly these enforced by monetary establishments and different high-value targets. This functionality considerably will increase the chance of success for unauthorized entry makes an attempt utilizing credentials harvested through infostealer logs, additional enhancing the post-exploitation worth of Lumma infections.”
The disclosures come as stealer malware like Vidar and Atomic macOS Stealer (AMOS) are being distributed utilizing the ClickFix technique through lures for the DeepSeek synthetic intelligence (AI) chatbot, in accordance with Zscaler ThreatLabz and eSentire.
Phishing assaults have additionally been noticed abusing a JavaScript obfuscation technique that makes use of invisible Unicode characters to characterize binary values, a method that was first documented in October 2024.
The strategy entails making use of Unicode filler characters, particularly Hangul half-width (U+FFA0) and Hangul full-width (U+3164), to characterize the binary values 0 and 1, respectively, and changing every ASCII character within the JavaScript payload to their Hangul equivalents.
“The assaults have been extremely customized, together with private info, and the preliminary JavaScript would attempt to invoke a debugger breakpoint if it have been being analyzed, detect a delay, after which abort the assault by redirecting to a benign web site,” Juniper Risk Labs stated.