From Phishing to Malware: AI Becomes Russia's New Cyber Weapon in War on Ukraine
Vulnerabilities mentionedAll →
| CVE | Vulnerability | CVSS | EPSS | Flags | Affected | Exposure | Published |
|---|---|---|---|---|---|---|---|
| CVE-2023-43770 | Persistent Cross-Site Scripting in Roundcube Webmail (Exploited in the Wild) Roundcube Webmail versions before 1.4.14, 1.5.x before 1.5.4, and 1.6.x before 1.6.3 contain a persistent cross-site scripting (XSS) flaw (CWE-79) caused by how program/lib/Roundcube/rcube_string_replacer.php converts plain text into clickable links. An attacker sends a text/plain email containing crafted links; when the recipient views the message, the crafted link text is turned into HTML that runs attacker-controlled script, which persists and executes in the victim's webmail session. Successful exploitation lets the attacker execute JavaScript with the victim's session, enabling mailbox access, theft of session credentials, and actions performed as the user (CVSS 6.1, scope-changed with limited confidentiality and integrity impact). Anyone running an affected Roundcube instance is exposed, including the roundcube package shipped with Debian Linux. The flaw is being actively exploited: CISA added it to the Known Exploited Vulnerabilities catalog on 2024-02-12, and EPSS assigns a 58.5% probability of exploitation within 30 days (99th percentile). Do: Upgrade Roundcube to 1.4.14, 1.5.4, or 1.6.3 (or later, per branch), or install the updated roundcube package on Debian. Because the bug is in CISA's KEV catalog, U.S. federal agencies must apply the vendor fix by the catalog due date, and other defenders should prioritize patching internet-facing webmail servers. Review webmail access logs for suspicious message views or account activity, and consider forcing session re-authentication for accounts that opened crafted plain-text messages. | 6.1 | 58% | KEV |
| massplausibly millions of users across tens of thousands of internet-exposed Roundcube instances (estimate) | |
| CVE-2024-27443 | Cross-Site Scripting in Zimbra Collaboration Suite CalendarInvite (Classic Webmail) CVE-2024-27443 is a cross-site scripting vulnerability (CWE-79) in the CalendarInvite feature of the Zimbra webmail classic user interface in Synacor Zimbra Collaboration Suite (ZCS). It is triggered when a user's browser renders an email containing a crafted calendar header, causing attacker-controlled JavaScript to execute within the webmail session. Successful exploitation allows an attacker to run arbitrary JavaScript in the victim's browser, enabling session/cookie theft and actions performed as the victim inside webmail; no public proof-of-concept is known and CVSS has not yet been scored. Organizations running ZCS where users access mail through the classic webmail UI are affected (deployments restricted to the modern UI are not impacted), though specific affected version ranges have not been published in the available data. The flaw was added to the CISA KEV catalog on 2025-05-19, confirming exploitation in the wild, and EPSS currently estimates a 23.6% probability of exploitation within 30 days (98th percentile). Do: Update ZCS to the patched release for your branch per Synacor/Zimbra's security advisory (specific fixed version numbers are not included in the available data) and confirm whether the classic webmail UI is enabled for any users. Review webmail access logs for suspicious calendar-invite traffic, and note that federal agencies must apply vendor mitigations per BOD 22-01 guidance or discontinue use of the product if mitigations are unavailable. | 6.1 | 24% | KEV |
| mass≈ mass | |
| CVE-2024-37383 | Cross-Site Scripting in Roundcube Webmail via SVG animate attributes CVE-2024-37383 is a cross-site scripting vulnerability (CWE-79) in Roundcube Webmail caused by insufficient handling of SVG 'animate' attributes when HTML email content is rendered. An attacker triggers it by sending a crafted HTML email containing a malicious SVG animate element; when the recipient views the message in Roundcube, attacker-controlled JavaScript executes in the context of the victim's webmail session. Successful exploitation allows theft of session cookies and credentials, access to mailbox contents, sending mail as the victim, or redirection to phishing pages, and has been used in campaigns that steal credentials and email. All Roundcube Webmail deployments before 1.5.7 and 1.6.x before 1.6.7 are affected, including Roundcube packages shipped with Debian; because the attack requires only viewing a malicious email, any exposed webmail user is a potential victim. The flaw is under active exploitation: unknown threat actors have used it in phishing campaigns, it carries an EPSS of 73.3%, and CISA added it to the Known Exploited Vulnerabilities catalog on 2024-10-24. Do: Upgrade Roundcube to version 1.5.7 or 1.6.7 (or later) immediately; Debian users should install the updated roundcube package from their repository. Organizations subject to CISA BOD 22-01 must apply vendor mitigations or discontinue use per the KEV entry. Review webmail logs for phishing emails containing SVG animate elements and investigate for credential theft or anomalous mailbox activity. | 6.1 | 73% | KEV |
| masslikely millions of webmail users across hundreds of thousands of deployed instances, with tens of thousands of instances internet-exposed | |
| CVE-2025-27915 | Stored XSS in Zimbra Collaboration Suite Classic Web Client via Malicious ICS Files CVE-2025-27915 is a stored cross-site scripting (XSS) flaw in the Classic Web Client of Synacor Zimbra Collaboration Suite (ZCS) 9.0, 10.0, and 10.1, caused by insufficient sanitization of HTML content in ICS calendar files. It is triggered when a user views an email message containing a malicious ICS entry, at which point embedded JavaScript executes in the victim's session via an ontoggle event handler inside a tag. A successful attacker can run arbitrary JavaScript in the victim's session and perform unauthorized actions on the account, notably creating email filters that silently redirect messages to attacker-controlled addresses, enabling data exfiltration. Any organization running the affected ZCS versions whose users read mail through the Classic Web Client is exposed, since delivery of a single crafted email can compromise a session. Exploitation is confirmed in the wild: the flaw was added to CISA KEV on 2025-10-07, and public reporting describes active zero-day attacks, including targeting of the Brazilian military via malicious ICS files. Do: Upgrade affected ZCS 9.0, 10.0, and 10.1 deployments to the latest patched builds per Zimbra's security advisory and apply any vendor-recommended mitigations (federal agencies must act per CISA KEV/BOD 22-01 requirements). Hunt for suspicious ICS-based emails and calendar entries, and review user mail filter rules for unauthorized forwarding or redirection to attacker-controlled addresses. Consider restricting or disabling Classic Web Client access until systems are patched. | 5.4 | 4% | KEV PoC |
| large≈30,000–50,000 internet-exposed Zimbra servers; total deployments including internal-only instances likely higher | |
| CVE-2025-49113 | Authenticated PHP Object Deserialization RCE in Roundcube Webmail Roundcube Webmail before 1.5.10 and 1.6.x before 1.6.11 contains a PHP object deserialization flaw (CWE-502) that allows remote code execution by authenticated users. The bug is in program/actions/settings/upload.php, where the _from URL parameter is not validated before deserialization, so any logged-in user can trigger it with a crafted URL to the settings upload action, with no user interaction required. Successful exploitation gives the attacker code execution on the web server with high confidentiality, integrity, and availability impact (CVSS 3.1 8.8). All deployments running affected versions are exposed, including Roundcube packages shipped with Debian Linux. The flaw reportedly existed for roughly a decade before disclosure, carries a 98.9% EPSS score (top percentile), and was added to CISA's KEV catalog on 2026-02-20, confirming exploitation in the wild. Do: Upgrade to Roundcube 1.6.11 or 1.5.10 (or later); Debian administrators should install the fixed roundcube package via security updates. Hunt for compromise by reviewing web access logs for requests to the settings upload action with unusual or crafted _from parameters, and check the web server runtime for unexpected processes, files, or webshells. Federal agencies must apply mitigations per vendor instructions under BOD 22-01 or discontinue use of the product if mitigations are unavailable. | 8.8 | 99% | KEV PoC ×2 |
| masslikely tens of thousands of internet-exposed Roundcube instances serving millions of mailboxes |
Indicators of compromiseauto-extracted · verify before use · export allAll →
| Type | Indicator | Context |
|---|---|---|
| domain | ipfs.io | flare Workers, Telegram, Telegra.ph, Teletype.in, Firebase, ipfs.io, mocky.io, to host malware or phishing pages, or turn them |
| domain | mocky.io | kers, Telegram, Telegra.ph, Teletype.in, Firebase, ipfs.io, mocky.io, to host malware or phishing pages, or turn them into a dat |
| domain | teletype.in | Drive, Bitbucket, Cloudflare Workers, Telegram, Telegra.ph, Teletype.in, Firebase, ipfs.io, mocky.io, to host malware or phishing p |
Full article568 words · extracted from thehackernews.com · click to collapse
Ravie LakshmananOct 09, 2025Artificial Intelligence / Malware
Russian hackers' adoption of artificial intelligence (AI) in cyber attacks against Ukraine has reached a new level in the first half of 2025 (H1 2025), the country's State Service for Special Communications and Information Protection (SSSCIP) said.
"Hackers now employ it not only to generate phishing messages, but some of the malware samples we have analyzed show clear signs of being generated with AI – and attackers are certainly not going to stop there," the agency said in a report published Wednesday.
SSSCIP said 3,018 cyber incidents were recorded during the time period, up from 2,575 in the second half of 2024 (H2 2024). Local authorities and military entities witnessed an increase in attacks compared to H2 2024, while those targeting government and energy sectors declined.
One notable attack observed involved UAC-0219's use of malware called WRECKSTEEL in attacks aimed at state administration bodies and critical infrastructure facilities in the country. There is evidence to suggest that the PowerShell data-stealing malware was developed using AI tools.
Some of the other campaigns registered against Ukraine are listed below -
- Phishing campaigns orchestrated by UAC-0218 targeting defense forces to deliver HOMESTEEL using booby-trapped RAR archives
- Phishing campaigns orchestrated by UAC-0226 targeting organizations involved in the development of innovations in the defense industrial sector, local government bodies, military units, and law enforcement agencies to distribute a stealer called GIFTEDCROOK
- Phishing campaigns orchestrated by UAC-0227 targeting local authorities, critical infrastructure facilities, and Territorial Recruitment and Social Support Centers (TRCs and SSCs) that leverage ClickFix-style tactics or SVG file attachments to distribute stealers like Amatera Stealer and Strela Stealer
- Phishing campaigns orchestrated by UAC-0125, a sub-cluster with ties to Sandworm, that sent email messages containing links to a website masquerading as ESET to deliver a C#-based backdoor named Kalambur (aka SUMBUR) under the guise of a threat removal program
SSSCIP said it also observed the Russia-linked APT28 (aka UAC-0001) actors weaponizing cross-site scripting flaws in Roundcube and (CVE-2023-43770, CVE-2024-37383, and CVE-2025-49113) and Zimbra (CVE-2024-27443 and CVE-2025-27915) webmail software to conduct zero-click attacks.
"When exploiting such vulnerabilities, attackers typically injected malicious code that, through the Roundcube or Zimbra API, gained access to credentials, contact lists, and configured filters to forward all emails to attacker-controlled mailboxes," SSSCIP said.
"Another method of stealing credentials using these vulnerabilities was to create hidden HTML blocks (visibility: hidden) with login and password input fields, where the attribute autocomplete='on' was set. This allowed the fields to be auto-filled with data stored in the browser, which was then exfiltrated."
The agency also revealed that Russia continues to engage in hybrid warfare, synchronizing its cyber operations in conjunction with kinetic attacks on the battlefield, with the Sandworm (UAC-0002) group targeting organizations in the energy, defense, internet service providers, and research sectors.
Furthermore, several threat groups targeting Ukraine have resorted to abusing legitimate services, such as Dropbox, Google Drive, OneDrive, Bitbucket, Cloudflare Workers, Telegram, Telegra.ph, Teletype.in, Firebase, ipfs.io, mocky.io, to host malware or phishing pages, or turn them into a data exfiltration channel.
"The use of legitimate online resources for malicious purposes is not a new tactic," SSSCIP said. "However, the number of such platforms exploited by Russian hackers has been steadily increasing in recent times."
Found this article interesting? Follow us on Google News, Twitter and LinkedIn to read more exclusive content we post.
Text extracted automatically; images, tables and formatting may be missing. Original: https://thehackernews.com/2025/10/from-phishing-to-malware-ai-becomes.html