閉める
閉める
明日に向けたネットワーク
明日に向けたネットワーク
サポートするアプリケーションとユーザー向けに設計された、より高速で、より安全で、回復力のあるネットワークへの道を計画します。
          Netskopeを体験しませんか?
          Netskopeプラットフォームを実際に体験する
          Netskope Oneのシングルクラウドプラットフォームを直接体験するチャンスです。自分のペースで進められるハンズオンラボにサインアップしたり、毎月のライブ製品デモに参加したり、Netskope Private Accessの無料試乗に参加したり、インストラクター主導のライブワークショップに参加したりできます。
            SSEのリーダー。 現在、シングルベンダーSASEのリーダーです。
            SSEのリーダー。 現在、シングルベンダーSASEのリーダーです。
            Netskope、2024年ガートナー、シングルベンダーSASEのマジック・クアドラントでリーダーの1社の位置付けと評価された理由をご確認ください。
              ダミーのためのジェネレーティブAIの保護
              ダミーのためのジェネレーティブAIの保護
              ジェネレーティブ AI の革新的な可能性と堅牢なデータ セキュリティ プラクティスのバランスを取る方法をご覧ください。
                ダミーのための最新のデータ損失防止(DLP)eBook
                最新の情報漏えい対策(DLP)for Dummies
                クラウド配信型 DLP に移行するためのヒントとコツをご紹介します。
                  SASEダミーのための最新のSD-WAN ブック
                  SASEダミーのための最新のSD-WAN
                  遊ぶのをやめる ネットワークアーキテクチャに追いつく
                    リスクがどこにあるかを理解する
                    Advanced Analytics は、セキュリティ運用チームがデータ主導のインサイトを適用してより優れたポリシーを実装する方法を変革します。 Advanced Analyticsを使用すると、傾向を特定し、懸念事項に的を絞って、データを使用してアクションを実行できます。
                        レガシーVPNを完全に置き換えるための6つの最も説得力のあるユースケース
                        レガシーVPNを完全に置き換えるための6つの最も説得力のあるユースケース
                        Netskope One Private Accessは、VPNを永久に廃止できる唯一のソリューションです。
                          Colgate-Palmoliveは、スマートで適応性のあるデータ保護により「知的財産」を保護します
                          Colgate-Palmoliveは、スマートで適応性のあるデータ保護により「知的財産」を保護します
                            Netskope GovCloud
                            NetskopeがFedRAMPの高認証を達成
                            政府機関の変革を加速するには、Netskope GovCloud を選択してください。
                              一緒に素晴らしいことをしましょう
                              Netskopeのパートナー中心の市場開拓戦略により、パートナーは企業のセキュリティを変革しながら、成長と収益性を最大化できます。
                                Netskopeソリューション
                                Netskope Cloud Exchange
                                Netskope Cloud Exchange(CE)は、セキュリティ体制全体で投資を活用するための強力な統合ツールをお客様に提供します。
                                  Netskopeテクニカルサポート
                                  Netskopeテクニカルサポート
                                  クラウドセキュリティ、ネットワーキング、仮想化、コンテンツ配信、ソフトウェア開発など、多様なバックグラウンドを持つ全世界にいる有資格のサポートエンジニアが、タイムリーで質の高い技術支援を行っています。
                                    Netskopeの動画
                                    Netskopeトレーニング
                                    Netskopeのトレーニングは、クラウドセキュリティのエキスパートになるためのステップアップに活用できます。Netskopeは、お客様のデジタルトランスフォーメーションの取り組みにおける安全確保、そしてクラウド、Web、プライベートアプリケーションを最大限に活用するためのお手伝いをいたします。

                                      Ousaban: LATAM Banking Malware Abusing Cloud Services

                                      Aug 04 2022

                                      Summary

                                      Ousaban (a.k.a. Javali) is a banking malware that emerged between 2017 and 2018, with the primary goal of stealing sensitive data from financial institutions in Brazil. This malware is developed in Delphi and it comes from a stream of LATAM banking trojans sourced from Brazil, sharing similarities with other families like Guildma, Casbaneiro, and Grandoreiro. Furthermore, the threat often abuses cloud services, such as Amazon S3 to download second stage payloads, and Google Docs to retrieve the C2 configuration.

                                      Netskope Threat Labs came across recent Ousaban samples that are abusing multiple cloud services throughout the attack flow, such as Amazon or Azure to download its payloads and log the victim’s IP, and Pastebin to retrieve the C2 configuration. The malware is downloaded through MSI files either by a JavaScript or a Delphi DLL, and is targeting more than 50 financial institutions in Brazil. Furthermore, we also found Telegram abuse in the malware code, likely used for C2 communication via Webhooks.

                                      Diagram of Ousaban attack flow

                                      In this blog post, we will analyze Ousaban, showing its delivery methods, obfuscation techniques, and C2 communication.

                                      Delivery methods

                                      Ousaban is delivered through malicious MSI files spread in phishing emails. In this campaign, we found that the MSI file downloads and executes the second stage either through JavaScript or a PE file.

                                      Delivery by JavaScript
                                      In the first scenario, the JavaScript is executed via CustomAction.

                                      Example of MSI file executing JavaScript.
                                      MSI file executing JavaScript.

                                      The JavaScript code is obfuscated, likely in an attempt to slow down analysis.

                                      Example of JavaScript code extracted from the MSI file
                                      JavaScript code extracted from the MSI file

                                      Looking at the deobfuscated code, these are the steps executed by the malware:

                                      1. Creates an empty file to be used as a flag in case the MSI is executed twice (similar concept as Mutex usage);
                                      2. Downloads the second stage from the cloud, either from Amazon or Azure;
                                      3. Decompress the ZIP file downloaded from the cloud and renames the main executable;
                                      4. Sends a simple GET request to another URL (Azure or another attacker-controlled server), alerting the attacker and logging the victim’s IP;
                                      5. Executes the main file via WMIC.
                                      Screenshot of deobfuscated JavaScript extracted from the MSI file.
                                      Deobfuscated JavaScript extracted from the MSI file.

                                      Delivery by File

                                      We also found Ousaban being delivered without JavaScript. In this case, we can see a file named “avisoProtesto.exe” being executed via MSI CustomAction.

                                      Example of MSI executing a PE file.
                                      MSI executing a PE file.

                                      “avisoProtesto.exe” is a signed and non-malicious binary exploited to execute the malicious DLL via DLL search order hijacking.

                                      Example of non-malicious binary used to load the malicious DLL.
                                      Non-malicious binary used to load the malicious DLL.

                                      This is possible because the non-malicious binary loads a DLL named “crashreport.dll” without specifying the real path of the library. Therefore, the attacker places a DLL with the same name in the same folder of the executable, making it load the malicious DLL instead.

                                      Example of Binary vulnerable to DLL hijacking.
                                      Binary vulnerable to DLL hijacking.

                                      In this case, both next-stage and tracker URL are loaded from a text file, named “FileLinks”.

                                      Screenshot of malicious URLs loaded by the malware.
                                      Malicious URLs loaded by the malware.

                                      All the files we analyzed were downloading the next stage from the cloud, either Amazon or Azure. In some cases, the URL used to log the victim’s IP address was also from Azure. All the URLs can be found in our GitHub repository.

                                      Loading the second stage

                                      The binary downloaded from the cloud is a ZIP file containing the next stage payload, which is a Delphi DLL executed by a non-malicious binary.

                                      Example of Files downloaded from the cloud.
                                      Files downloaded from the cloud.

                                      The file executed by the malware is a non-malicious executable with a valid signature (“Securityo6Z3.exe”).

                                      Screenshot of certificate found in the file executed by the malware.
                                      Certificate found in the file executed by the malware.

                                      The malicious DLL is then loaded by the non-malicious binary through a DLL search order hijacking vulnerability, the same technique that is used by some of the downloaders.

                                      Example of non-malicious binary loading the next stage DLL.
                                      Non-malicious binary loading the next stage DLL.

                                      Second stage

                                      The second stage is a Delphi malware responsible for decrypting and loading Ousaban’s payload in the following flow:

                                      1. Loads the encrypted bytes of Ousaban from disk;
                                      2. Decrypts Ousaban payload using a key stored in the “.data” section;
                                      3. Decrypts the code that runs Ousaban using the same key, stored in the “.data” section.
                                      Example of decryption key and encrypted code stored in the “.data” section of the second stage
                                      Decryption key and encrypted code stored in the “.data” section of the second stage.

                                      The encrypted payload of Ousaban is located among the files downloaded from the cloud, named “ZapfDingbats.pdf”.

                                      Example of third stage encrypted among files downloaded from the cloud.
                                      Third stage encrypted among files downloaded from the cloud.

                                      Once running, the second stage loads Ousaban’s encrypted bytes, which will be decrypted using the key stored in the PE “.data” section.

                                      Screenshot of encrypted Ousaban payload being loaded.
                                      Encrypted Ousaban payload being loaded.

                                      Aside from decrypting the payload, the second stage also decrypts the code that will execute Ousaban in runtime, probably to slow down reverse engineering.

                                      Screenshot of second stage decrypting and loading Ousaban payload.
                                      Second stage decrypting and loading Ousaban payload.

                                      We created a Python script that can be used to statically decrypt Ousaban payloads, using the same algorithm found in the malware. The code can be found in our GitHub repository.

                                      Important API calls used by this stage are also dynamically resolved, another common technique to slow down reverse engineering.

                                      Screenshot of APIs dynamically loaded by the malware.
                                      APIs dynamically loaded by the malware.

                                      Ousaban payload

                                      Ousaban is a Delphi banking trojan, mainly focused on stealing sensitive data from financial institutions in Brazil. As previously mentioned, Ousaban shares many similarities with other Brazilian banking malware, such as the algorithm to decrypt the strings and overlay capabilities.

                                      Ousaban commonly packs/protects its payloads with UPX or Enigma.

                                      Example of Ousaban payload packed with UPX.
                                      Ousaban payload packed with UPX.

                                      One of the most characteristic aspects of Brazilian-sourced banking malware is the algorithm used to encrypt/decrypt important strings.

                                      Screenshot of Ousaban encrypted strings.
                                      Ousaban encrypted strings.

                                      The algorithm used as a base by these trojans was originally demonstrated in a Brazilian magazine called “Mestres Da Espionagem Digital” in 2008. Simply put, it parses the hexadecimal string and uses a chained XOR operation with the key and the previous byte of the string.

                                      Screenshot of part of the algorithm to decrypt the strings, commonly found in Brazilian banking malware.
                                      Part of the algorithm to decrypt the strings, commonly found in Brazilian banking malware.

                                      We created a Python script that can be used to decrypt strings from malware that uses this algorithm, such as Ousaban, Guildma, Grandoreiro, and others. The code can be used to decrypt a single string:

                                      Screenshot of decrypting a single string from the malware.
                                      Decrypting a single string from the malware.

                                      Or to decrypt multiple strings at once, saving the result in a JSON file and also providing the option to show in the console.

                                      Screenshot of decrypting multiple strings from the malware.
                                      Decrypting multiple strings from the malware.

                                      Like other Brazilian-sourced malware, Ousaban monitors the title text from the active window and compares it with a list of strings, to verify if the victim is accessing the website or an application of one of its targets.

                                      Exammple of malware monitoring windows titles.
                                      Malware monitoring windows titles.

                                      In the files we analyzed, we found Ousaban targeting over 50 different financial institutions. If the window title matches one of the targets, Ousaban starts the communication with the C2 address, providing the option to the attacker to access the machine remotely.

                                      C2 communication

                                      Ousaban stores the C2 address remotely. In this case, the malware is using Pastebin to fetch the data. In 2021, this malware was also spotted using Google Docs to fetch this information.

                                      Within the files downloaded from the cloud by the first stage, there’s a file named “Host”, which stores the external location of the C2 configuration. The information is encrypted with the same algorithm used in the strings.

                                      Example of C2 configuration stored on Pastebin.
                                      C2 configuration stored on Pastebin.

                                      The data is stored in a dictionary, where the C2 host is also encrypted with the same algorithm used in the strings.

                                      ""
                                      Retrieving and decrypting the C2 server address.

                                      Ousaban only starts the communication once a targeted company is identified.

                                      Screenshot of Ousaban C2 communication.
                                      Ousaban C2 communication.

                                      Lastly, the Ousaban samples we analyzed contain a routine to communicate via Telegram using Webhooks, likely to be used as a secondary channel.

                                      Screenshot of part of Ousaban code to communicate via Telegram.
                                      Part of Ousaban code to communicate via Telegram.

                                      Conclusion

                                      Ousaban is a malware designed to steal sensitive information from several financial institutions, mainly based in Brazil. Ousaban shares many similarities with other Brazilian-based banking trojans, such as Guildma and Grandoreiro. Also, as we demonstrated in this analysis, the attackers behind this threat are abusing multiple cloud services throughout the attack chain. We believe that the use of the cloud will continue to grow among attackers especially due to cost and ease.

                                      Protection

                                      Netskope Threat Labs is actively monitoring this campaign and has ensured coverage for all known threat indicators and payloads. 

                                      • Netskope Threat Protection
                                        • Win32.Malware.Heuristic
                                        • Win32.Infostealer.Heuristic
                                      • Netskope Advanced Threat Protection provides proactive coverage against this threat.
                                        • Gen.Malware.Detect.By.StHeur indicates a sample that was detected using static analysis
                                        • Gen.Malware.Detect.By.Sandbox indicates a sample that was detected by our cloud sandbox

                                      IOCs

                                      All the IOCs related to this campaign and scripts can be found in our GitHub repository.

                                      author image
                                      Gustavo Palazolo
                                      Gustavo Palazolo is an expert in malware analysis, reverse engineering and security research, working many years in projects related to electronic fraud protection.
                                      Gustavo Palazolo is an expert in malware analysis, reverse engineering and security research, working many years in projects related to electronic fraud protection.

                                      Stay informed!

                                      Subscribe for the latest from the Netskope Blog