Pierluigi Paganini
August 30, 2024
Google TAG (Threat Analysis Group) researchers observed the Russia-linked group APT29 (aka SVR group, BlueBravo, Cozy Bear, Nobelium, Midnight Blizzard, and The Dukes). using exploits previously used by surveillance software vendors NSO Group and Intellexa. The circumstance suggests that the nation-state actors may have purchased the surveillance tools from the surveillance firms.
TAG experts detected multiple exploit campaigns between November 2023 and July 2024 that were used in watering hole attacks on Mongolian government websites. These campaigns targeted both iOS and Android users, exploiting iOS versions older than 16.6.1 and Android Chrome versions from m121 to m123. Despite the vulnerabilities being addressed, Google notified Apple, Android, and Chrome teams, as well as the Mongolian CERT, to address the compromised sites. The findings highlight the ongoing risk posed by watering hole attacks and the collaboration between nation-state actors and commercial surveillance vendors.
APT29 compromised the websites of the Mongolian cabinet (cabinet.gov.mn) and Ministry of Foreign Affairs (mfa.gov.mn), they embedded hidden iframes linked to attacker-controlled sites. In November 2023, these iframes led to the site track-adv.com, delivering a CVE-2023-41993 exploit targeting iPhones running iOS 16.6.1 or older, deploying a cookie-stealing payload previously associated with APT29. The researchers noticed that by February 2024, mfa.gov.mn was compromised again, this time linking to ceo-adviser.com and delivering the same exploit and payload, but updated to target specific websites like webmail.mfa.gov.mn. In July 2024, the site mfa.gov.mn was compromised a third time, redirecting Android Chrome users to track-adv.com to exploit CVE-2024-5274 and CVE-2024-4671, deploying a Chrome information-stealing payload.
Google also observed Apple Safari campaigns that were conducted between November 2023 and February 2024. Threat actors conducted a watering hole attack to deliver an iOS exploit through CVE-2023-41993. When accessed by an iPhone or iPad, the compromised sites used an iframe to deploy a reconnaissance payload that validated the device before downloading a WebKit exploit to steal browser cookies. This exploit only affected devices running iOS 16.6.1 or older, while those on iOS 16.7 or with lockdown mode enabled were protected.
“When visited with an iPhone or iPad device, the watering hole sites used an iframe to serve a reconnaissance payload, which performed validation checks before ultimately downloading and deploying another payload with the WebKit exploit to exfiltrate browser cookies from the device.” reads the TAG’s report. “The WebKit exploit did not affect users running the current iOS version at the time (iOS 16.7), working only on iOS versions 16.6.1 or older. Users with lockdown mode enabled were not affected even when running a vulnerable iOS version.”
The exploit employed in this campaign shared the exact same trigger as one previously used by Intellexa, suggesting a strong connection between the authors or providers of both exploits. However, it is unclear how the attackers obtained this exploit.
“The iOS exploit loaded the same cookie stealer framework that TAG observed in March 2021 when a Russian government-backed attacker exploited CVE-2021-1879 to acquire authentication cookies from prominent websites such as LinkedIn, Gmail and Facebook.” continues the report. “In that campaign, attackers used LinkedIn Messaging to target government officials from western European countries by sending them malicious links.”
Google TAG also shared details of a new watering hole attack that occurred in late July 2024. Threat actors compromised the mfa.gov.mn website, targeting Android users via a Google Chrome exploit chain. This attack was similar to a previous iOS exploit, both aiming to steal credential cookies using n-day vulnerabilities. However, the Chrome attack involved more technical complexity, including a sandbox escape to bypass Chrome’s site isolation. APT29 used obfuscated JavaScript to inject a malicious iframe and employed ECDH key exchange for encryption. The attack also utilized indexedDB for storing client-side status information, with unique identifiers used consistently throughout the stages.
APT29 chained two vulnerabilities in the exploit used in this attack. The first, CVE-2024-5274, was used to compromise the Chrome browser’s renderer. This vulnerability was discovered and reported as an in-the-wild 0-day in May 2024 by Google’s Threat Analysis Group (TAG) and Chrome Security, after being used by the NSO Group, a commercial surveillance vendor (CSV). Although the attackers adapted NSO Group’s exploit, their version targeted only Chrome versions 121 to 123, whereas the NSO exploit supported a broader range of versions, from 107 to 124. The experts pointed out that despite sharing a similar trigger, the two exploits are conceptually different.
“What is clear is that APT actors are using n-day exploits that were originally used as 0-days by CSVs. It should be noted that outside of common exploit usage, the recent watering hole campaigns otherwise differed in their approaches to delivery and second-stage objectives.” Google TAG concludes.
Follow me on Twitter: @securityaffairs and Facebook and Mastodon
(SecurityAffairs – hacking, APT29)
