Machine learning–powered Android Trojans bypass script-based Ad Click detection
嗯,用户让我帮忙总结一下这篇文章的内容,控制在一百个字以内,而且不需要用“文章内容总结”或者“这篇文章”这样的开头。直接写描述即可。 首先,我需要通读整篇文章,理解其主要内容。文章讲的是一个新的Android点击欺诈木马家族,使用TensorFlow.js机器学习模型来视觉检测和点击广告,绕过了传统的基于脚本的点击方法。这个恶意软件通过隐藏的WebView运行,并远程加载模型。研究人员发现它通过Xiaomi的应用商店和其他第三方网站传播,影响了多个游戏应用,并且这些木马可以将设备变成僵尸网络的一部分,进行DDoS攻击、窃取个人信息等。 接下来,我需要提取关键信息:新的Android木马家族、使用TensorFlow ML、绕过传统检测、通过隐藏WebView运行、远程加载模型、传播途径包括Xiaomi和第三方网站、影响多个游戏应用、功能包括广告点击、DDoS攻击、窃取信息等。 然后,我要把这些信息浓缩到100字以内。可能的结构是:介绍木马家族及其技术手段,传播方式,以及带来的威胁。 最后,确保语言简洁明了,不使用复杂的术语,同时涵盖所有关键点。 </think> 新的Android点击欺诈木马家族利用TensorFlow ML视觉检测广告并自动点击,绕过传统脚本方法。该恶意软件通过隐藏WebView运行,并从远程服务器加载模型。研究人员发现其通过Xiaomi应用商店及第三方网站传播,并嵌入多个游戏应用中。该木马可将设备变为僵尸网络节点,用于DDoS攻击及窃取个人信息。 2026-1-22 19:19:39 Author: securityaffairs.com(查看原文) 阅读量:2 收藏

Machine learning–powered Android Trojans bypass script-based Ad Click detection

A new Android click-fraud trojan family uses TensorFlow ML to visually detect and tap ads, bypassing traditional script-based click techniques.

Researchers at cybersecurity firm Dr.Web discovered a new Android click-fraud trojan family that uses TensorFlow.js ML models to visually detect and tap ads, avoiding traditional script-based methods. The malware is distributed via Xiaomi’s GetApps, it runs in a hidden “phantom” WebView and loads models remotely. The malicious code analyzes screenshots, and mimics real user behavior, making it more resilient against dynamic ads.

In phantom mode, Android.Phantom.2.origin uses a hidden WebView browser that, on command from the playstations[.]click server, loads a target site and a “phantom” JavaScript file containing ad-automation logic and TensorFlowJS. A trained ML model is downloaded from a remote server, screenshots of a virtual screen are analyzed, and detected ad elements are automatically clicked. In signaling mode, the Trojan uses WebRTC, with dllpgd[.]click acting as a signaling server, to stream a live video of the virtual browser to attackers, who can remotely control it by clicking, scrolling, and entering text.

The new Android clicker Trojan family is controlled by the dllpgd[.]click server. Several popular games from a single developer were initially clean but later updated with the Android.Phantom.2.origin Trojan.

The researchers identified several mobile games containing malware:

  • Creation Magic World (over 32,000 downloads);
  • Cute Pet House (>34,000 downloads);
  • Amazing Unicorn Party (>13,000 downloads);
  • Sakura Dream Academy (>4,000 downloads);
  • Theft Auto Mafia (>61,000 downloads);
  • Open World Gangsters (>11,000 downloads).

“All infected games are hosted by a single developer, SHENZHEN RUIREN NETWORK CO., LTD., and the Trojans are embedded within them and run alongside the apps.” reads the report published by Dr Web.

“The initial versions of the games were free of malware. On September 28/29, the developer released updates for the games, which contained the Android.Phantom.2.origin Trojan . It operates in two modes, which are referred to in the program code as signaling and phantom.”

On October 15/16, the games were updated again to include Android.Phantom.5, a dropper carrying the Android.Phantom.4.origin remote loader, which downloads additional click-fraud Trojans. These simpler modules rely on JavaScript click scripts rather than ML or video streams. Android.Phantom.5 also enables downloading a required WebRTC library, expanding capabilities.

Beyond Xiaomi apps, the malware spreads via third-party APK sites like Moddroid and Apkmody through modified Spotify, YouTube, Deezer, and Netflix apps.

The researchers reported that many “Editor’s Choice” apps on Moddroid are infected. Malicious APKs also circulate on Telegram channels and a Discord server with 24,000 users promoting an infected Spotify X app.

The server data shows infections across multiple languages, with Spanish, French, German, Polish, and Italian most affected after English; many Asian languages lacked dedicated chats. These trojans can turn devices into bots for DDoS attacks, enable illegal activity, drain battery and data, and leak personal information via spyware modules. Users without updated antivirus protection are especially at risk, with children and those seeking unofficial app access being particularly vulnerable.

“We recommend avoiding downloading mods from untrusted websites and channels. Verifying the sources of mods or apps typically requires time, experience, and a keen eye.” concludes the report.

Follow me on Twitter: @securityaffairs and Facebook and Mastodon

Pierluigi Paganini

(SecurityAffairs – hacking, TensorFlow ML)




文章来源: https://securityaffairs.com/187215/malware/machine-learning-powered-android-trojans-bypass-script-based-ad-click-detection.html
如有侵权请联系:admin#unsafe.sh