Siri, Apple’s voice activated personal assistant and knowledge navigator application, can be leveraged to steal sensitive information from iOS smartphones in a stealthy manner, according to researchers.
Luca Caviglione of the National Research Council of Italy and Wojciech Mazurczy of the Warsaw University of Technology warn that malicious actors could use Siri for stealthy data exfiltration by using a method that’s based on steganography, the practice of hiding information.
Malware that uses such techniques is becoming more and more common. For example, the developers of Duqu, Alureon and Zeus have all used innocent-looking image files for data transmission and command and control (C&C) communications. Other threats have used the traffic generated by popular applications such as Skype to hide data.
iOS malware is also increasingly common, but many of these threats are not very stealthy. However, Mazurczy and Caviglione have demonstrated that iOS malware could become difficult to detect.
When users talk to Siri, their voice is processed with the Speex Codec, and the data is transmitted to Apple’s servers where the voice input is translated to text.
The attack method developed by the researchers, dubbed iStegSiri, involves controlling the “shape” of this traffic to embed sensitive data from the device. This covert channel could be used to send credit card numbers, Apple IDs, passwords, and other sensitive information from the phone to the criminal mastermind, researchers said in their paper.
An iStegSiri attack takes place in three phases. In the first phase, the secret message is converted into an audio sequence based on voice and silence alternation. Then, the sound pattern is provided to Siri as input through the internal microphone. Finally, the recipient of the secret message inspects the traffic going to Apple’s servers and extracts the information based on a decoding scheme. Secret messages are highlighted by a specific set of features, the experts noted.
In their experiments, Mazurczy and Caviglione managed to use this method to exfiltrate data at a rate of 0.5 bytes per second. At this speed, it would take roughly 2 minutes to send a 16-digit payment card number to the attacker.
The iStegSiri attack can be effective because it doesn’t require the installation of additional software components and it doesn’t need the device’s alteration. On the other hand, it only works on jailbroken devices and attackers somehow need to be able to intercept the modified Siri traffic. However, the researchers highlighted that the purpose of iStegSiri is to help the security community with the detection of malware on the iOS platform.
“Because information-hiding methods use very specific technological traits, no current off-the-shelf products effectively detect covert communications. This forces security experts to craft dedicated countermeasures for each method,” researchers wrote in their paper. “With iStegSiri, the ideal countermeasure acts on the server side. For example, Apple should analyze patterns within the recognized text to determine if the sequence of words deviates significantly from the used language’s typical behaviors. Accordingly, the connection could be dropped to limit the covert communication’s data rate. “
The researchers told IEEE Spectrum that they haven’t made specific details on iStegSiri public to prevent cybercriminals from leveraging their work.

Eduard Kovacs (@EduardKovacs) is a contributing editor at SecurityWeek. He worked as a high school IT teacher for two years before starting a career in journalism as Softpedia’s security news reporter. Eduard holds a bachelor’s degree in industrial informatics and a master’s degree in computer techniques applied in electrical engineering.
More from Eduard Kovacs
- Unpatched Security Flaws Expose Water Pump Controllers to Remote Hacker Attacks
- 3CX Confirms Supply Chain Attack as Researchers Uncover Mac Component
- OpenSSL 1.1.1 Nears End of Life: Security Updates Only Until September 2023
- Google Links More iOS, Android Zero-Day Exploits to Spyware Vendors
- ChatGPT Data Breach Confirmed as Security Firm Warns of Vulnerable Component Exploitation
- Thousands Access Fake DDoS-for-Hire Websites Set Up by UK Police
- Intel Boasts Attack Surface Reduction With New 13th Gen Core vPro Platform
- Dole Says Employee Information Compromised in Ransomware Attack
Latest News
- Anti-Bot Software Firm DataDome Banks $42M Financing
- Unpatched Security Flaws Expose Water Pump Controllers to Remote Hacker Attacks
- 500k Impacted by Data Breach at Debt Buyer NCB
- Chinese Cyberspies Use ‘Melofee’ Linux Malware for Stealthy Attacks
- Why Endpoint Resilience Matters
- Microsoft Cloud Vulnerability Led to Bing Search Hijacking, Exposure of Office 365 Data
- 3CX Confirms Supply Chain Attack as Researchers Uncover Mac Component
- UK Introduces Mass Surveillance With Online Safety Bill
