APT29
Also known as: atk7, blue kitsune, bluebravo, cloaked ursa, cozy bear, cozyduke, dark halo, g0016, grizzly steppe, group 100, ice relic, icecap
Profile
APT29 (also known as Cozy Bear, The Dukes, Nobelium, Midnight Blizzard or, under Google Threat Intelligence's new naming, ICE RELIC) is a cyber-espionage group that the US and UK governments attribute to Russia's Foreign Intelligence Service (SVR), active since at least 2008. It mainly targets governments, diplomacy, think tanks, NGOs, defence, energy, research institutes and technology providers in Europe, Ukraine, the US and NATO countries. It is known for the SolarWinds supply-chain compromise (2020), mass exploitation of JetBrains TeamCity (CVE-2023-42793), a 2024 watering-hole campaign against Mongolian government websites documented by Google, and the 2024 access to Microsoft's corporate email via password spraying, an incident confirmed by the company. It stands out for its mastery of cloud and identity environments (OAuth application abuse, token theft, device-code phishing, impersonation of technology providers); in 2024-2025 it sent thousands of emails with signed RDP files to more than 100 organisations, used fake wine-tasting invitations to deliver GrapeLoader and WineLoader to European embassies, and in 2025 Amazon disrupted one of its watering-hole campaigns. It remains highly active in 2026: since May a sub-cluster Microsoft tracks as Storm-2945 has compromised Wi-Fi captive portals at hotels, conference centres and airports (the CaptiveCrunch campaign) to steal Microsoft 365 credentials and, through fake updates, install the CornFlake RAT and the ChocoShell infostealer, managed from the FruitStone panel; according to Anthropic, the group used its Claude AI model to rebuild already-detected malware and sustain espionage against more than 20 government, diplomatic, defence and think-tank organisations in Ukraine and Europe; and, according to Sekoia (with medium-to-high confidence, an assessment not yet backed by any government or CERT), in the second half of 2026 it ran the 'Poisoned Source' watering-hole campaign against iOS and Android users, using compromised legitimate sites (including those of a Ukrainian gas producer, Ukrainian and Bulgarian military-related sites and a US chemical industry trade association), fake domains mirroring real websites and, between 27 August and 3 September, a tampered script from a Ukrainian web analytics service loaded by thousands of sites, to deliver n-day exploits (for already-patched flaws) without user interaction and steal cookies, messaging databases and credentials; the payloads established no persistence and the group may have obtained the exploits from a third party.
Active since: 2008
Initial access
- Spearphishing with links or attachments (ZIP with DLL side-loading, signed RDP files, .lnk shortcuts launching PowerShell payloads)
- Device-code phishing and abuse of legitimate Microsoft authentication flows
- Password spraying against accounts without MFA and legacy test accounts
- Exploitation of JetBrains TeamCity (CVE-2023-42793) and other exposed services
- Software supply-chain compromise (SolarWinds, 2020; tampered script of a Ukrainian web analytics and SEO tracking service, 27 August to 3 September 2026)
- Watering hole on compromised legitimate websites with injected profiling JavaScript delivering platform-specific, zero-click n-day exploit chains for iOS and Android
- Fake domains mirroring and proxying real or high-profile sites (e.g. a US 250th anniversary initiative) and injecting malicious scripts
- Fake sites impersonating a London luxury hotel, possibly for spearphishing against diplomatic delegations
- Compromised Wi-Fi captive portals (hotels, conferences, airports) that intercept connectivity checks and serve fake browser or OS updates and ClickFix-style prompts
Tools
SUNBURST, TEARDROP, WineLoader, GrapeLoader, CornFlake (RAT), ChocoShell (infostealer), FruitStone (C2 panel), DarkSword (iOS and Android exploit kit), AndroSnitch (Android exploit chain), UltimateFingerprinter (visitor profiling script), Cobalt Strike, Mimikatz, PowerSploit, Empire, AADInternals and OAuth application abuse, Claude (Anthropic), abused to rebuild detected malware
What defenders should watch
- Require phishing-resistant MFA on all accounts, including test and legacy accounts, and restrict or block the device-code flow where not needed.
- Regularly audit OAuth applications, granted permissions, device registrations and Microsoft 365 sign-ins, especially for staff returning from travel.
- Block .rdp and .lnk files received by email, monitor PowerShell executions launched from shortcuts and outbound RDP connections, and train travelling staff not to install updates offered by Wi-Fi portals or paste commands a website asks for (ClickFix), using the corporate VPN on public networks.
- Keep iOS, Android and mobile browsers (especially Chrome) on corporate devices up to date: the mobile campaign exploits already-fixed vulnerabilities without user interaction (CVE-2024-27834, listed in CISA's KEV catalogue, CVE-2025-14174, CVE-2025-27038, CVE-2025-43510, CVE-2025-43520, CVE-2025-43529, CVE-2026-20700 and CVE-2026-3910), so up-to-date devices are not affected.
- Check the integrity of third-party scripts (analytics, SEO tracking) loaded by your own websites and search DNS and proxy logs for the published campaign domains, such as ad.starx-net[.]com (exfiltration) or theowo[.]club; quickly patch exposed build and development servers such as TeamCity, and do not rely on signatures alone, because the group uses AI to rebuild detected malware.
Most affected countries
No data yet.
Most affected sectors
No data yet.
Indicators of compromise 13
IPs, domains, URLs, hashes and files attributed to this group by the sources
Extracted from the public reports and news we read about this group, with the source for each one. Check them against your own telemetry; an indicator may be reused or expired.
| Indicator | Context | First seen | Last seen | Source |
|---|---|---|---|---|
Domain ad.starx-net.com | dominio de la campaña para revisar en DNS y proxy | Nov 4, 2026 | Nov 4, 2026 | Orbitaley |
Domain ads-it-assets.com | dominio de la campaña para revisar en DNS y proxy | Nov 4, 2026 | Nov 4, 2026 | Orbitaley |
Domain assets-int.com | dominio de la campaña para revisar en DNS y proxy | Nov 4, 2026 | Nov 4, 2026 | Orbitaley |
Domain cdn.adv-static.com | dominio de la campaña para revisar en DNS y proxy | Nov 4, 2026 | Nov 4, 2026 | Orbitaley |
Domain edge-ord.clouds-stacks.com | dominio de la campaña para revisar en DNS y proxy | Nov 4, 2026 | Nov 4, 2026 | Orbitaley |
Domain static.flux-telemetry.com | dominio de la campaña para revisar en DNS y proxy | Nov 4, 2026 | Nov 4, 2026 | Orbitaley |
Domain theowo.club | dominio de la campaña para revisar en DNS y proxy | Nov 4, 2026 | Nov 4, 2026 | Orbitaley ×2 |
Domain wpvortex-mod.com | dominio de la campaña para revisar en DNS y proxy | Nov 4, 2026 | Nov 4, 2026 | Orbitaley |
IP 109.104.106.107 | infraestructura de la campaña para bloquear | Nov 4, 2026 | Nov 4, 2026 | Orbitaley |
IP 188.190.17.232 | infraestructura de la campaña para bloquear | Nov 4, 2026 | Nov 4, 2026 | Orbitaley |
IP 192.121.46.25 | infraestructura de la campaña para bloquear | Nov 4, 2026 | Nov 4, 2026 | Orbitaley |
Domain america250.holiday | dominio falso que imita sitio de alto perfil | Nov 4, 2026 | Nov 4, 2026 | The Hacker News ×2 |
URL http://ad.starx-net.com:8881/telemetry/v12 | endpoint de exfiltración de datos | Nov 4, 2026 | Nov 4, 2026 | The Hacker News |
Vulnerabilities attributed
| CVE | Severity | EPSS | Active exploitation |
|---|---|---|---|
| CVE-2024-27834 | Medium (5.5) | 1.5% | ⚠ Active exploitation |
| CVE-2025-14174 | High (8.8) | 43% | ⚠ Active exploitation |
| CVE-2025-27038 | High (7.5) | 1.0% | ⚠ Active exploitation |
| CVE-2025-43510 | High (7.8) | 0.37% | ⚠ Active exploitation |
| CVE-2025-43520 | Medium (5.5) | 0.44% | ⚠ Active exploitation |
| CVE-2025-43529 | High (8.8) | 18% | ⚠ Active exploitation |
| CVE-2026-20700 | High (7.8) | 1.7% | ⚠ Active exploitation |
| CVE-2026-3910 | High (8.8) | 2.7% | ⚠ Active exploitation |
Priority mitigations
MITRE ATT&CK mitigations that cover the most techniques used by this group.
- Privileged Account Management (46 techniques covered)
- User Account Management (37 techniques covered)
- Audit (32 techniques covered)
- Network Intrusion Prevention (27 techniques covered)
- Multi-factor Authentication (24 techniques covered)
- Execution Prevention (24 techniques covered)
- Disable or Remove Feature or Program (23 techniques covered)
- User Training (22 techniques covered)
MITRE ATT&CK techniques
Techniques marked ✦ are not attributed to this group by MITRE ATT&CK: Orbitaley infers them from public reports (a tool it uses, a vulnerability it exploits or, by AI, its documented behaviour). The source is shown in brackets.
Recent victims
No data yet.
Claims are published by the criminal groups themselves and are unverified until the organisation or the press confirms them. Names of natural persons (sole traders, individual professionals) are anonymised under the GDPR. We never link to leak sites or stolen data. To request the removal or anonymisation of an entry, contact the site administrator.
Sources: RansomLook (CC BY 4.0), MITRE ATT&CK®, MISP Galaxy, Google News.