CriticalCVE-2024-27834CVE-2025-43529CVE-2026-20700CVE-2025-14174CVE-2025-43510CVE-2025-43520+2 CVEs 4 Hunt LeadsIOCsRemote AttackerHands-On-KeyboardAPT29DarkSwordAndroSnitchUltimateFingerprinterSekoia

APT29 Poisoned an Analytics Script to Hack Phones Across Ukraine

Sekoia says APT29 hijacked the Plerdy analytics script, hacked sites in four countries and faked the America250 and a London hotel site to push iOS and Android n-day chains, including DarkSword and Apple's newly KEV-listed CVE-2024-27834.

Russia’s APT29 poisoned a Ukrainian web analytics script for a week this summer, quietly loading a phone-exploit kit onto thousands of mostly Ukrainian websites, according to new research from Sekoia. The same operation hacked high-profile sites in the US, Ukraine, the Czech Republic and Bulgaria and ran lookalike copies of the America250 site and a London luxury hotel, all to steal cookies, Wi-Fi passwords and in some cases a phone’s entire contents. Every exploit was already patched; one of them, Apple’s CVE-2024-27834, landed on CISA’s exploited list on Nov. 4.

For defenders

Attribution: Sekoia assesses with moderate to high confidence that this is APT29 (Midnight Blizzard, Cozy Bear, The Dukes), widely attributed to Russia’s SVR, or one of its subgroups. Google TAG told Sekoia the framework resembles APT29’s 2024 watering holes on Mongolian government sites; a national CERT matched leftover artifacts on one victim site to a 2025 APT29 campaign.

Three ways in: JavaScript injected into compromised sites (a Ukrainian gas producer, a US chemical-industry trade body, a Bulgarian military-related site, a far-right Czech movement); reverse-proxy lookalikes america250[.]holiday and theowo[.]club that patched a malicious script into the real pages (T1583.001 (Acquire Infrastructure: Domains)); and, from Aug. 27 to Sept. 3, 2026, a compromise of the Plerdy SEO tracker’s main.js, loaded by thousands of sites including Ukrainian army, special forces and drone-equipment sites (T1195.002 (Supply Chain Compromise: Compromise Software Supply Chain), T1189 (Drive-by Compromise)).

The chain: every vector loads 5b4c8fea5cbcee273e8d830f52-temfp.js, a fingerprinting script the operators call UltimateFingerprinter, padded with a 2.3 MB AngularJS copy. It posts 394 device fields to the delivery host, which answers with exploit JavaScript for real phones (T1592 (Gather Victim Host Information)). Sekoia pulled three chains, all n-days (T1203 (Exploitation for Client Execution)):

Chain Targets CVEs Takes
A iOS 16.7 (works 16.0 to 17.4) CVE-2025-43529, CVE-2024-27834 (PAC bypass) Cookies from running apps
B (DarkSword) iOS 18.5 CVE-2025-43529, CVE-2026-20700, CVE-2025-14174, CVE-2025-43510, CVE-2025-43520 SMS/iMessage, mail, WhatsApp and Telegram databases, Keychain, contacts, calls, location, Health, photos, iCloud Drive
C (AndroSnitch) Android Chrome 139 to 145, not 143 CVE-2026-3910, CVE-2025-27038 (Adreno) Chrome cookies, logins, history; saved Wi-Fi passwords

Everything goes to ad.starx-net[.]com:8881/telemetry/v12 (T1041 (Exfiltration Over C2 Channel)); nothing persists. CVE-2024-27834 is on CISA KEV since Nov. 4, federal due date Nov. 18 (fixed in iOS/iPadOS 16.7.8 and 17.5, Safari 17.5, macOS Sonoma 14.5). Sekoia withheld some hosts and its YARA rules (TLP:AMBER for CERTs). We covered DarkSword’s earlier web lures in Dating and AI-companion lures load iPhone exploit chains.

Network
static.flux-telemetry[.]com (delivery, 109.104.106[.]107)
edge-ord.clouds-stacks[.]com (delivery load balancer, 95.133.231[.]28)
ad.starx-net[.]com (exfiltration, 195.58.140[.]105, port 8881)
adsprism-stat[.]com (delivery)
ads-it-assets[.]com
cdn.adv-static[.]com
wpvortex-mod[.]com
assets-int[.]com
america250[.]holiday (lookalike reverse proxy)
theowo[.]club (lookalike reverse proxy)

Host
5b4c8fea5cbcee273e8d830f52-temfp.js (UltimateFingerprinter)
VDGH_TOKEN_REF (iOS chain marker cookie)
SHA256 387a7fb768db1350a9788d9704e7a72abe07152734e0ff026af72b417e36268c (AndroSnitch)

What it means for you

This attack needed nothing from victims except an out-of-date phone and a visit to the wrong website. Sekoia is clear that every bug used was already fixed, so an updated iPhone or Android phone was safe. The people most at risk are those connected to Ukraine’s military, government and defense, and to diplomatic events, but the Plerdy compromise put ordinary visitors of thousands of Ukrainian sites in range too. Stolen cookies can let an attacker into your webmail without your password.

Do this now

  1. Update your phone now: current iOS or iPadOS, and the latest Chrome and Android security patch. If your phone can’t get updates anymore, replace it.
  2. If you work in or with Ukrainian defense or government, sign out of all sessions on your webmail and key accounts and sign back in, so any stolen cookies stop working.
  3. On iPhones of people who may be targeted, consider turning on Lockdown Mode.
  4. Website owners using Plerdy or other third-party scripts: check your pages for unexpected script loads, and add a Content Security Policy that limits where scripts can load from.

Source article: Poisoned Source: APT29 watering holes target mobile devices with n-day exploits
Publisher: Sekoia Threat Detection & Research (Amaury Garçon, Félix Aime, Robin Kwiatkowski, TDR Team)
Published: November 4, 2026
URL: https://www.sekoia.com/blog/poisoned-source-apt29-watering-holes-targets-mobile-devices-with-n-day-exploits

Executive Summary

Sekoia TDR attributes, with moderate to high confidence, a large watering-hole campaign uncovered in mid-October 2026 to APT29 (also tracked as Midnight Blizzard, Cozy Bear and The Dukes), the intrusion set widely attributed to Russia’s Foreign Intelligence Service (SVR), or to one of its subgroups. The operators used three delivery vectors: JavaScript injected into the templates or existing script files of compromised legitimate websites in Ukraine, the US, the Czech Republic and Bulgaria; attacker-registered lookalike domains (america250[.]holiday, theowo[.]club) acting as reverse proxies in front of the real America250 and Raffles London at The OWO sites and patching a script tag into each page; and a one-week supply-chain compromise (Aug. 27 to Sept. 3, 2026) of the Ukrainian Plerdy SEO tracker script, loaded by thousands of mostly Ukrainian sites including defense and military-recruitment sites. Every vector loads the same fingerprinting script, 5b4c8fea5cbcee273e8d830f52-temfp.js (“UltimateFingerprinter”), hidden behind a 2.3 MB AngularJS copy, which posts a 394-field device profile to static.flux-telemetry[.]com and executes whatever exploit JavaScript comes back; delivery is refused only for implausible screen sizes or non-touch devices. Sekoia recovered three n-day chains: Class A for iOS 16.7 (CVE-2025-43529 WebKit RCE plus the CVE-2024-27834 PAC bypass) stealing app cookies; Class B, the DarkSword kit for iOS 18.5 (CVE-2025-43529, CVE-2026-20700, CVE-2025-14174, CVE-2025-43510, CVE-2025-43520), harvesting messages, mail, WhatsApp and Telegram databases, Keychain, location, Health, photos and iCloud Drive; and Class C for Android Chrome 139 to 145 except 143 (CVE-2026-3910 plus the Adreno driver bug CVE-2025-27038) dropping AndroSnitch to take Chrome cookies, logins and history and saved Wi-Fi passwords. All three exfiltrate to ad.starx-net[.]com:8881/telemetry/v12 without persistence. The objective is intelligence collection, with stolen session cookies enabling access to webmail and other browser-based services. CISA added CVE-2024-27834 to its Known Exploited Vulnerabilities catalog on Nov. 4, 2026. Hunt leads, false positive expectations and indicators are detailed below; Sekoia withheld its YARA rules and some hosts (TLP:AMBER for CERTs and law enforcement), so no published rule content is reproduced.

ATT&CK chain: Resource Development → Initial Access → Execution → Privilege Escalation → Discovery → Collection → Exfiltration

Hunt Lead Grid

Hunt Lead Level Hunt Lead Name ATT&CK Risk of False Positives
1 (Junior) Delivery and Exfiltration Infrastructure Sweep T1189 (Drive-by Compromise), T1041 (Exfiltration Over C2 Channel) Low. The ten domains are attacker-registered and the exfiltration endpoint uses an unusual port and path; the three IPs sit with hosting providers that serve other customers, so an IP-only hit needs a matching domain, port or time window before it counts.
1 (Junior) Fingerprinting Script Injected Into Websites You Run T1608.004 (Stage Capabilities: Drive-by Target), T1195.002 (Supply Chain Compromise: Compromise Software Supply Chain) Low. The script file name is fixed across all victims and the delivery domains are unique to this campaign; the only legitimate match is the real Plerdy tracker URL, so the rule anchors on the injected reference, not the Plerdy include itself.
2 (Pro) Mobile Browser Fingerprint POST, Then Script, Then Port 8881 T1592 (Gather Victim Host Information), T1203 (Exploitation for Client Execution) Medium. Analytics and fraud-prevention services also POST large browser fingerprints, so the lead anchors on a POST to an /api/<32-character token>/v1 path on a newly registered domain followed by HTTPS to port 8881 from the same mobile client.
3 (Elite) Webmail Session Reuse After a Mobile Cookie Theft T1539 (Steal Web Session Cookie), T1550.004 (Use Alternate Authentication Material: Web Session Cookie) Medium to High. Users roam between networks and devices all the time; the lead needs a session token first seen on a mobile client that later reached this infrastructure, then reused from a different ASN or device fingerprint without a fresh sign-in.

Delivery and Exfiltration Infrastructure Sweep

Sekoia publishes the delivery host static.flux-telemetry[.]com (registered April 2, 2026, 109.104.106[.]107, AS395092 Shock Hosting), the Plerdy-stage load balancer edge-ord.clouds-stacks[.]com (registered Aug. 26, 2026, 95.133.231[.]28, AS219399 servinga), the exfiltration host ad.starx-net[.]com (registered April 3, 2026, 195.58.140[.]105, AS395092), the additional delivery domains adsprism-stat[.]com, ads-it-assets[.]com, cdn.adv-static[.]com, wpvortex-mod[.]com and assets-int[.]com, and the reverse-proxy lookalikes america250[.]holiday (registered May 19, 2026) and theowo[.]club (registered June 18, 2026). Victims’ browsers fetch /module/5b4c8fea5cbcee273e8d830f52-temfp.js from a delivery host, POST JSON to /api/<nonce>/v1, and compromised phones send stolen data over HTTPS to port 8881 at /telemetry/v12. The detection logic is a retrospective sweep of DNS, secure web gateway, firewall and mobile-network (MDM or mobile threat defense) logs for any of the ten domains or three IPs, with priority on any connection to ad.starx-net[.]com or to TCP 8881 on 195.58.140[.]105, which indicates successful exploitation rather than a mere visit.

Use DNS resolver logs, proxy logs, firewall and NetFlow, and mobile threat defense or MDM network telemetry, since most hits will come from phones on corporate Wi-Fi or a managed VPN. Sweep back to April 2026, when the core domains were registered, and pay particular attention to Aug. 27 to Sept. 3, 2026 for organizations whose staff browse Ukrainian sites. Scope first to staff working on Ukraine, defense, diplomacy, government relations and political parties, and to anyone who attended or organized events tied to America250 or stayed near Whitehall. Block the domains and the exfiltration IP, but do not block the shared hosting ASNs wholesale. A hit on the exfiltration endpoint means the device was compromised: escalate to incident response, assume that phone’s browser and app sessions, saved Wi-Fi credentials and, for iOS 18.5, message and Keychain contents are in attacker hands, and revoke sessions and rotate credentials accordingly. Because Sekoia withheld some hosts, a clean sweep is not proof of non-exposure.

Hunt Lead Level: 1 (Junior Analyst)

ATT&CK: Initial Access → T1189 (Drive-by Compromise); Exfiltration → T1041 (Exfiltration Over C2 Channel)

Sources: “Poisoned Source: APT29 watering holes target mobile devices with n-day exploits”, Sekoia TDR, Nov. 4, 2026, https://www.sekoia.com/blog/poisoned-source-apt29-watering-holes-targets-mobile-devices-with-n-day-exploits (source article)

Fingerprinting Script Injected Into Websites You Run

On directly compromised sites the operators either edited the main page template to append a script element after page load, or appended the same code to an existing legitimate JavaScript library file (as on a Bulgarian military-related site), so the page HTML can look untouched while a bundled .js file carries the injection. In the Plerdy stage the tampered file was the tracker itself, hxxps://a.plerdy[.]com/public/js/click/main.js, which loaded the legitimate tracker plus a script from edge-ord.clouds-stacks[.]com. In every case the end target is a file named 5b4c8fea5cbcee273e8d830f52-temfp.js served from a /module/ path on an attacker domain. The detection logic is a content-integrity check across every web property and CMS you operate: search rendered pages, template files and all served JavaScript for the string temfp.js, for any of the delivery domains, and for dynamically created script elements pointing at domains not in your approved third-party list; for Plerdy customers, compare the main.js your pages received during Aug. 27 to Sept. 3, 2026 (from CDN, WAF or browser-side CSP reports) against the vendor’s clean version.

Use web server and CDN access logs, file-integrity monitoring on web roots, CMS change history, and Content-Security-Policy violation reports if you collect them. Sweep at least six months, because the core domains were live from April and Sekoia found artifacts from an earlier 2025 APT29 campaign still sitting in one victim’s files. Scope to public-facing sites likely to draw government, defense, diplomatic or political visitors, but include marketing and event microsites, which are often less watched. Allowlisting the Plerdy include itself is fine; the anchor is any extra script reference it or your templates pull in. Removing the injected line is not enough: treat a hit as a server compromise, find the original access path, rotate CMS and hosting credentials, and notify your national CERT, since your visitors may have been exploited. A strict CSP with an explicit script-src allowlist generalizes to this and most other script-injection watering holes.

Hunt Lead Level: 1 (Junior Analyst)

ATT&CK: Resource Development → T1608.004 (Stage Capabilities: Drive-by Target); Initial Access → T1195.002 (Supply Chain Compromise: Compromise Software Supply Chain)

Sources: “Poisoned Source: APT29 watering holes target mobile devices with n-day exploits”, Sekoia TDR, Nov. 4, 2026, https://www.sekoia.com/blog/poisoned-source-apt29-watering-holes-targets-mobile-devices-with-n-day-exploits (source article)

Mobile Browser Fingerprint POST, Then Script, Then Port 8881

The delivery sequence is consistent across all three chains. A mobile browser loads the roughly 2.3 MB temfp.js, then sends an HTTPS POST with Content-Type application/json, a large body of device fields, to /api/ followed by a 32-character alphanumeric token and /v1 on the same delivery host; the response is JavaScript that the page injects and runs. On iOS the exploit first sets a cookie named VDGH_TOKEN_REF as a replay guard. For Class B, further stages are fetched from hxxps://ad.starx-net[.]com/api/<nonce>/v1; for every chain, collection ends with an HTTPS POST to port 8881 at /telemetry/v12, and on Android AndroSnitch sends it via libcurl with Filename and X-Device-Id headers. The detection logic, built on proxy or TLS-inspection logs, flags a client with a mobile Safari or Android Chrome user agent that, within a few minutes, downloads a JavaScript file over 2 MB from a domain first seen in the last 90 days, POSTs JSON to a path matching /api/[A-Za-z0-9]{32}/v1 on that domain, and then opens an HTTPS session to a non-standard port such as 8881 on any host, with extra weight when the client runs iOS 16.0 to 17.4, iOS 18.5 or Chrome 139 to 145.

Use secure web gateway or TLS-inspecting proxy logs for corporate Wi-Fi and managed VPN traffic, joined to MDM inventory for OS and browser versions; where you can’t inspect TLS, fall back to SNI, destination port and newly-observed-domain feeds. Sweep 180 days. Scope to managed and BYOD phones of high-risk staff first. Expect false positives from ad-tech and anti-fraud fingerprinting, which also POSTs JSON blobs; suppress established vendors by registered domain age and reputation rather than by path, and require the port-8881 or non-standard-port follow-on for an alert. Blocking outbound 8881 at the perimeter is a reasonable hardening step but will not help a phone on a home or mobile network, so pair this lead with enforced minimum OS versions in MDM. Treat any full-sequence match as a likely compromise and escalate; the pattern generalizes to other fingerprint-gated mobile exploit kits that reuse commercial surveillance tooling.

Hunt Lead Level: 2 (Pro)

ATT&CK: Reconnaissance → T1592 (Gather Victim Host Information); Execution → T1203 (Exploitation for Client Execution)

Sources: “Poisoned Source: APT29 watering holes target mobile devices with n-day exploits”, Sekoia TDR, Nov. 4, 2026, https://www.sekoia.com/blog/poisoned-source-apt29-watering-holes-targets-mobile-devices-with-n-day-exploits (source article)

The payoff of all three chains is session material. Class A walks running iOS apps’ sandbox containers and reads every Cookies.binarycookies file it can reach; Class C copies Chrome’s Cookies, Login Data and History databases on Android; Class B takes the Keychain, account credentials and Safari data along with everything else. Sekoia notes this lets the operators reuse authenticated sessions to reach browser-based services such as webmail, and Google TAG linked APT29’s 2024 campaign to the same cookie-theft goal. The detection logic correlates identity-provider and webmail sign-in logs with the infrastructure hits from the first lead: for each user whose phone reached any listed domain or port 8881, look for subsequent use of an existing session (no interactive sign-in, no MFA event) from a new ASN, country, device ID or user agent, especially hosting or VPN ASNs, and for mailbox actions from that session such as bulk mail access, new forwarding rules or new app passwords.

Use identity-provider sign-in and token logs, webmail or Microsoft 365 / Google Workspace audit logs, and the network telemetry from the first lead. Sweep from April 2026 forward. Scope to users with a positive infrastructure hit, then widen to all high-risk users, because Sekoia withheld some delivery and exfiltration hosts and a clean network sweep does not rule out exposure. Expect noise from travel and VPN use; baseline each user’s normal ASNs and devices, and raise confidence when the session reuse follows a hit within days. Resetting the password alone is the wrong response, since a stolen cookie survives it: revoke all sessions and refresh tokens, re-register MFA from a known-clean device, and review mailbox rules. Token binding or device-bound session credentials, where your provider supports them, defeat this class of theft generally.

Hunt Lead Level: 3 (Elite)

ATT&CK: Credential Access → T1539 (Steal Web Session Cookie); Lateral Movement / Defense Evasion → T1550.004 (Use Alternate Authentication Material: Web Session Cookie)

Sources: “Poisoned Source: APT29 watering holes target mobile devices with n-day exploits”, Sekoia TDR, Nov. 4, 2026, https://www.sekoia.com/blog/poisoned-source-apt29-watering-holes-targets-mobile-devices-with-n-day-exploits (source article); “State-backed attackers and commercial surveillance vendors repeatedly use the same exploits”, Google TAG, Aug. 2024, https://blog.google/threat-analysis-group/state-backed-attackers-and-commercial-surveillance-vendors-repeatedly-use-the-same-exploits/ (additional source, APT29 2024 cookie-theft watering holes)

Hash Appendix

Exploit chains and AndroSnitch

SHA-256s come from the source article’s IOC table and chain sections; Sekoia published no MD5s or SHA-1s, and OTX AlienVault and VirusTotal were not checked. Most of these are JavaScript exploit stages served to browsers, so they are best matched in proxy content inspection or web caches rather than on endpoints.

MD5s
(not available)  # Class-A-Stage1 -> iOS 16.7 chain stage 1 (JS)
(not available)  # Class-A-Stage2 -> iOS 16.7 exploit + cookie stealer (JS, 101K)
(not available)  # Class-B-Stage1 -> DarkSword iOS 18.5 chain stage 1
(not available)  # Class-B-Stage2 -> DarkSword iOS 18.5 chain stage 2
(not available)  # Class-B-Stage3 -> DarkSword iOS 18.5 chain stage 3
(not available)  # Class-B component -> listed in Class B section
(not available)  # Class-B component 2 -> listed in Class B section
(not available)  # Class-C-Stage1-JS -> Android Chrome RCE (JS)
(not available)  # Class-C-Stage2-JS -> Android Chrome stage 2 (JS)
(not available)  # Class-C-Stage2-CVE-2025-27038 -> Adreno sandbox escape shellcode
(not available)  # Class-C component -> listed in Class C section
(not available)  # Class-C component 2 -> listed in Class C section
(not available)  # AndroSnitch -> libexploit.so cookie and Wi-Fi stealer
SHA1s
(not available)  # Class-A-Stage1 -> iOS 16.7 chain stage 1 (JS)
(not available)  # Class-A-Stage2 -> iOS 16.7 exploit + cookie stealer (JS, 101K)
(not available)  # Class-B-Stage1 -> DarkSword iOS 18.5 chain stage 1
(not available)  # Class-B-Stage2 -> DarkSword iOS 18.5 chain stage 2
(not available)  # Class-B-Stage3 -> DarkSword iOS 18.5 chain stage 3
(not available)  # Class-B component -> listed in Class B section
(not available)  # Class-B component 2 -> listed in Class B section
(not available)  # Class-C-Stage1-JS -> Android Chrome RCE (JS)
(not available)  # Class-C-Stage2-JS -> Android Chrome stage 2 (JS)
(not available)  # Class-C-Stage2-CVE-2025-27038 -> Adreno sandbox escape shellcode
(not available)  # Class-C component -> listed in Class C section
(not available)  # Class-C component 2 -> listed in Class C section
(not available)  # AndroSnitch -> libexploit.so cookie and Wi-Fi stealer
SHA256s
a7d2493df00565c21587b0cad2a9fe7ca1e5590acf796da6e499be0ffde44cc3  # Class-A-Stage1 -> iOS 16.7 chain stage 1 (JS)
bee3e014e3df87ddce0250b0985b87f6868a427a7b650a74b7ffedb9d57f3088  # Class-A-Stage2 -> iOS 16.7 exploit + cookie stealer (JS, 101K)
c170fda024183645c0aa681149304e50506509c3498bb6b9a5f5ced6aa73e5fa  # Class-B-Stage1 -> DarkSword iOS 18.5 chain stage 1
2ffb35e3f9e55286a930fb887e73764a913166648135fe0ca7551736d97d7bfa  # Class-B-Stage2 -> DarkSword iOS 18.5 chain stage 2
69b6cb327a4d5da99399718f6a66f84de69f89425b0a31f2dcee0682f811596f  # Class-B-Stage3 -> DarkSword iOS 18.5 chain stage 3
35c0cd19a7334b544c193e392bcd9a8f1d1f43cf6b61be5bf1d9e35537aff2ef  # Class-B component -> listed in Class B section
621112712ec515e8791129c62feda8fff6654a0116882401609503cacc08054a  # Class-B component 2 -> listed in Class B section
29d283ebd08d0cd3b2e6f4a67718d3d6e7a6be01d0613d716bce218dba57a928  # Class-C-Stage1-JS -> Android Chrome RCE (JS)
06c17b31d36838f3d29044784aa047d5c2d9174d3356c1211cf25f9df677445f  # Class-C-Stage2-JS -> Android Chrome stage 2 (JS)
183d62d682aea8018491c8559544e8d6a9ab0e1abd05ecbb6c97daa4f438d2b8  # Class-C-Stage2-CVE-2025-27038 -> Adreno sandbox escape shellcode
ae3241d96cc2d99538c7aff60f382cc8dac13859bcbea3ee84992d24d22a8108  # Class-C component -> listed in Class C section
41e6fd33a30d36dabff9ea408d9766a6583edcc107968f6dc1499f2586381967  # Class-C component 2 -> listed in Class C section
387a7fb768db1350a9788d9704e7a72abe07152734e0ff026af72b417e36268c  # AndroSnitch -> libexploit.so cookie and Wi-Fi stealer

References Appendix

  • “Poisoned Source: APT29 watering holes target mobile devices with n-day exploits”, Sekoia TDR (Amaury Garçon, Félix Aime, Robin Kwiatkowski), Nov. 4, 2026, https://www.sekoia.com/blog/poisoned-source-apt29-watering-holes-targets-mobile-devices-with-n-day-exploits (source article; vectors, chains, infrastructure, hashes, attribution)
  • “State-backed attackers and commercial surveillance vendors repeatedly use the same exploits”, Google Threat Analysis Group, Aug. 2024, https://blog.google/threat-analysis-group/state-backed-attackers-and-commercial-surveillance-vendors-repeatedly-use-the-same-exploits/ (APT29’s 2024 cookie-theft watering holes, cited by the source)
  • CISA Known Exploited Vulnerabilities Catalog, CVE-2024-27834 (added Nov. 4, 2026, due Nov. 18, 2026), https://www.cisa.gov/known-exploited-vulnerabilities-catalog
  • MITRE ATT&CK Enterprise technique pages linked above, MITRE, https://attack.mitre.org/

CVE at a glance

CVE-2024-27834MediumKEVApple Multiple Products
CVSS
5.5 Medium (v3.1)
EPSS
1.4% chance of exploitation in the next 30 days (higher than 72.8% of CVEs)
Exploited in the wild
Yes, on CISA’s KEV list since Nov 4, 2026
Public exploit
None reported in public exploit databases

Full CVE details

CVE-2025-43529HighKEVPublic exploitApple Multiple Products
CVSS
8.8 High (v3.1)
EPSS
25.7% chance of exploitation in the next 30 days (higher than 97.9% of CVEs)
Exploited in the wild
Yes, on CISA’s KEV list since Dec 15, 2025
Public exploit
Yes, reported in 7 GitHub repositories

Full CVE details

CVE-2026-20700HighKEVPublic exploitApple Multiple Products
CVSS
7.8 High (v3.1)
EPSS
1.7% chance of exploitation in the next 30 days (higher than 77% of CVEs)
Exploited in the wild
Yes, on CISA’s KEV list since Feb 12, 2026
Public exploit
Yes, reported in 2 GitHub repositories

Full CVE details

CVE-2025-14174HighKEVPublic exploitGoogle Chromium
CVSS
8.8 High (v3.1)
EPSS
55.3% chance of exploitation in the next 30 days (higher than 99% of CVEs)
Exploited in the wild
Yes, on CISA’s KEV list since Dec 12, 2025
Public exploit
Yes, reported in 3 GitHub repositories

Full CVE details

CVE-2025-43510HighKEVApple Multiple Products
CVSS
7.8 High (v3.1)
EPSS
0.4% chance of exploitation in the next 30 days (higher than 29.4% of CVEs)
Exploited in the wild
Yes, on CISA’s KEV list since Mar 20, 2026
Public exploit
None reported in public exploit databases

Full CVE details

CVE-2025-43520MediumKEVPublic exploitApple Multiple Products
CVSS
5.5 Medium (v3.1)
EPSS
0.4% chance of exploitation in the next 30 days (higher than 36.9% of CVEs)
Exploited in the wild
Yes, on CISA’s KEV list since Mar 20, 2026
Public exploit
Yes, reported in 1 GitHub repository

Full CVE details

Data from NVD, FIRST EPSS and CISA KEV, refreshed automatically. We report whether public exploit code exists but never link to it: fake proof-of-concepts that infect researchers are common.

More from the Roundup