Loading story…
Security researchers warned that attackers are increasingly weaponizing SVG files by embedding JavaScript and other active content into files that are often treated as harmless images. Recent campaigns used malicious SVG attachments and links for phishing, credential harvesting, malware delivery, and browser-based abuse, including fake voicemail lures, self-contained phishing pages rendered directly from the SVG, and redirectors that sent victims to attacker-controlled domains.
Analysis of recent samples found SVGs capable of obfuscated script execution, DOM manipulation, data exfiltration, and other in-browser actions that can bypass traditional defenses focused on conventional executables. Researchers linked one SVG-delivered script to a WordPress browser brute-force abuse pattern previously observed in 2024, while 2026 activity more commonly relied on redirects and embedded phishing content. Defenders were urged to treat SVGs as active executable content, inspect them for embedded scripts, and strengthen email and web filtering; published indicators included three domains and 22 SHA-1 hashes tied to the activity.
Get the infrastructure and lures behind it.
3 events from the most recent confirmed update back to the earliest known activity.
Gurucul published a high-severity threat research notice on malicious SVG abuse and provided indicators of compromise including three domains, 22 SHA-1 hashes, and detection queries. The notice recommended treating SVGs as executable content and strengthening email and web filtering.
ReversingLabs described an increase in malicious use of SVG files in early 2026, highlighting campaigns that used SVGs as redirectors, self-contained phishing pages, and script-enabled browser attack content. The analysis included fake voicemail-themed spearphishing attachments and SVGs that rendered full webpages while exfiltrating user input.
ReversingLabs reported that one analyzed script delivered via an SVG was associated with a 2024 browser brute-force abuse pattern targeting WordPress. The reference does not provide a specific date for when that 2024 activity occurred.
Vulnerabilities, threat actors, malware, products, organizations, breaches, and observables Mallory has linked to this story. Indicator values are masked here and available in full in the app.
Indicator values are masked on this page. View all 25 in Mallory Domains, IPs, hashes, and URLs are exportable to your SIEM.
Get the infrastructure, lures, and IOCs behind this campaign, ready to push into your email and identity stack.
2 references tracked. Mallory keeps watching after this page renders.
Map indicators from this story to your assets and identify affected systems in minutes.
Every observed campaign, victim, and pivot linked to actors named in this story.
Malware, exploits, and IOCs connected to the activity described here.
YARA, Sigma, and Snort rules deployed to your SIEM as soon as they’re published.
Get matching new stories delivered to your team as they break — not the next morning.
Ask questions about this story and take action on the answers.
Other stories Mallory is tracking that overlap on entities, sources, or topic labels.
Microsoft Threat Intelligence identified a sophisticated phishing campaign targeting U.S.-based organizations, leveraging artificial intelligence to craft highly convincing and evasive attacks. The campaign was detected in late August 2025 and involved the use of compromised small business email accounts to distribute phishing emails. These emails masqueraded as legitimate file-sharing notifications, enticing recipients to open an attachment that appeared to be a PDF but was actually a Scalable Vector Graphics (SVG) file. The attackers exploited the SVG format's flexibility, embedding dynamic and interactive code that could bypass many traditional security tools. Instead of using standard obfuscation techniques, the malicious payload was hidden within the SVG file by encoding it with a sequence of business-related terms, such as 'revenue,' 'operations,' and 'risk,' making the file appear as standard business data. This approach allowed the attackers to disguise the true intent of the file, which was to redirect users to a fake sign-in page designed to harvest credentials. Microsoft noted that the code within the SVG was likely generated or enhanced using large language models (LLMs), a form of AI that can produce complex, human-like text and code, further complicating detection efforts. The phishing emails also employed a self-addressed tactic, where the sender and recipient addresses matched, with actual targets hidden in the BCC field to evade basic detection heuristics. SVG files are particularly attractive to attackers because they are text-based, scriptable, and support features like invisible elements, encoded attributes, and delayed script execution, all of which can be used to sidestep static analysis and security scanning. Microsoft responded by using its own AI-powered security tools to analyze and block the campaign, highlighting the growing trend of AI versus AI in cybersecurity defense and offense. The campaign underscores the increasing adoption of AI by threat actors to automate the creation of convincing phishing lures and obfuscated malware. The use of compromised business email accounts added legitimacy to the phishing attempts, increasing the likelihood of successful credential theft. The incident demonstrates the evolving threat landscape, where attackers continuously adapt their methods to outpace security defenses. Organizations are advised to enhance their email security protocols, educate users about the risks of unexpected file attachments, and deploy advanced detection tools capable of analyzing complex file formats like SVG. The campaign serves as a warning that traditional security measures may be insufficient against AI-driven threats, necessitating a proactive and adaptive security posture. Microsoft’s public disclosure of the attack provides valuable intelligence for defenders to recognize and mitigate similar threats in the future. The incident also highlights the importance of monitoring for unusual email behaviors, such as self-addressed messages and unexpected file types. As AI becomes more accessible to both attackers and defenders, the arms race in cybersecurity is expected to intensify, with increasingly sophisticated attacks targeting organizations of all sizes.
Jun 29, 2026
Hackers have launched a sophisticated phishing campaign that impersonates official notices from the National Police of Ukraine to target organizations running Microsoft Windows. The campaign begins with emails containing malicious Scalable Vector Graphics (SVG) files, which are exploited due to their text-based nature that allows embedding of harmful scripts. When recipients open the SVG attachment, they are presented with a deceptive loading screen, after which the system downloads password-protected ZIP archives such as ergosystem.zip or smtpB.zip. The password is provided in the email to increase the appearance of legitimacy. Inside these archives, a Compiled HTML Help (CHM) file acts as the main trigger, executing a malicious script known as CountLoader. This loader connects to a remote server, exfiltrates basic system information, and then delivers the final malware payloads directly into memory, making detection challenging. The two primary payloads identified are Amatera Stealer, which is designed to steal sensitive data, and PureMiner, a cryptominer that hijacks system resources for illicit cryptocurrency mining. PureMiner is delivered using DLL sideloading techniques from the ZIP archive, while Amatera Stealer is deployed through the same infection chain. Security researchers have highlighted the dual threat posed by this campaign: data theft and resource hijacking. The use of SVG files is particularly concerning because they can contain embedded JavaScript and HTML, and on Windows systems, they are often opened by Microsoft Edge, which may lack adequate security controls if not the user's primary browser. The attackers use formal legal language in the phishing emails to pressure recipients into compliance, increasing the likelihood of infection. The campaign is notable for its fileless approach, as the malware is loaded directly into memory rather than being written to disk. This method complicates detection and remediation efforts for security teams. The infection chain leverages social engineering, technical obfuscation, and multiple stages to maximize its effectiveness. Researchers have also observed creative obfuscation within the SVG files, such as disguising malicious code with food and recipe-related variable names. The campaign demonstrates a high level of technical sophistication and adaptability, exploiting both human and technical vulnerabilities. Organizations are advised to educate users about the risks of opening unexpected attachments, especially SVG files, and to implement robust endpoint protection capable of detecting fileless threats. The campaign underscores the ongoing evolution of phishing tactics and the need for continuous vigilance in email security. Security teams should review their email filtering and endpoint monitoring policies to address this emerging threat vector. The use of password-protected archives and legitimate-seeming legal notices increases the campaign's success rate, making it a significant concern for enterprises.
Jun 29, 2026
A large-scale phishing campaign sent **26,589** fake voicemail emails to more than **5,500 organizations** between June 1 and August 4, using malicious SVG attachments to evade email security controls. The broadly distributed, non-targeted messages impersonated internal senders, personalized subject lines with recipients’ email local parts, and were largely delivered in weekday waves. The attachments were disguised as voicemail notifications and declared with the anomalous `text/plain` MIME type despite containing SVG/XML content. Embedded, obfuscated JavaScript used deferred execution and runtime script injection to retrieve remote content. Native Microsoft spam filtering assigned 75% of the messages Spam Confidence Level 0 or 1, classifying them as non-spam or low-confidence spam and allowing the campaign to bypass many standard defenses.
Aug 28, 2026