TITLE: A study of privacy of voice virtual assistant in Siri
The refinements of Artificial Intelligent applications have led to increasingly widespread of Virtual Assistants (VAs); smart-homes and smart phones have risen the need for potent encryption to save personal privacy. According to European Commission (2018), most users say that the “usefulness when hand/vision occupied” may become the major cause of talking to their devices. Overall, 30% of them mention that utterance is faster than inspecting, 24% are facing some issues not enabling them to type, and 12% find VAs easier than complex menu steps. However, one of the famed VAs named Siri, released by Apple, has a number of limitations, in which privacy is the prime issue. Siri can access messages either texts or emails, calendars, contact lists, read notes and other personal information freely. Chung et al. (n.d.) assert that VAs record 7/24 and listen to conversations and saved them in computing clouds which could be accessed by anyone. This has stimulated recent researchers to discuss the point that VAs invade users’ privacy. Thus, this report aims to highlight the problem of privacy in Siri, evaluate a solution which is given by Apple and discuss different issues of privacy.
This section will provide information on privacy, Virtual Assistant, Siri and some of its limitations.
Although privacy is an essential aspect of day-to-day, it might be difficult to control it in the digital world. According to Wacks (2015), privacy is worrying to secure sensitive data. In the 21st century, there was a common usage of social media sites, which led to a widespread concern about privacy in the digital age. The Electronic Communications Privacy Act of 1986, which was released before the evolution of the internet and the advent of social media, is implemented in The United States as a law to control online privacy (Purdy et al., 2013). Therefore, digital information has been grossly threatened and the introduction of strict regulation may alleviate this issue.
If you need assistance with writing your essay, our professional essay writing service is here to help!
Virtual assistants have been defined in recent research and explained as going through three stages. VA is a software that is carrying out tasks to help computer systems’ users by using their voices. Users ask virtual assistants and they respond similarly to natural interactions. Basically, this software contains three important parts, namely the speech-to-text engine (STT), the logic-handling engine and the text-to-speech engine (TTS) as presented in Figure 1.
Figure 1. Virtual assistant workflow (Pant, n.d.).
The first stage is the Speech-to-text engine (STT) as the name given, STT transforms a user’s oral message into written text that logic engine(s) can process it. This process includes: recording voice, scanning words and excluding any noise or distortion. After that, STT transforms the record into written text by natural language processing (NLP). Next stage is the Logic engine, in which words are received, handled, processed and passed outputs to the TTS. Finally, the text-to-speech engine (TTS) that transforms the output which is received to speech interactively with the user (Pant, n.d.).
In February 2010, Siri was a separate application (app) in Apple Store developed by Siri Inc. Siri Inc.’s co-founders —Dag Kittlaus, Adam Cheyer and Tom Gruber worked for the Defence Department on the Cognitive Assistant That Learns and Organizes (CALO) project before improving this technology. Interestingly, the manager of Apple and the chief executive officer of Apple made a great deal to own Siri. In 2011, Apple launched Siri on its iPhone 4S, as a virtual assistant, assisting users in doing their tasks, such as, reading and replying to texts. Soon after, Siri has been enhanced and updated since that time (Orner, 2016). However, during this period Apple launched these updates to fix security issues (Reisinger, 2017).
Despite the fact that VAs are beneficial and delightful, they also have a number of issues. The major problem of the voice virtual assistant is security; devices can be asked for doing tasks include sensitive information such as read calendar and emails. In one reported situation, a man explores that Siri in iPad would open the door for anyone standing outside the home and asking, “Siri open the door”. This has influenced Siri developers to introduce voice recognition feature to recognize user’s voice (Hoy, 2018).
Due to lack of research on the phases of developing Siri, this section will highlight explicit aspects of Siri.
Contacting with devices by utterance instead of a keyboard to accomplish tasks was a miracle, therefore, virtual assistant programs occupy a crucial place in recent electronic devices (see Appendix 1). In fact, the first step of enhancing voice virtual assistant was to overcome the limitation in vocabulary, but the ability of the software to comprehensive the variety of speech inputs became dilemma and demanded not only utterance recognition, but also the context, natural language understanding. In addition, the efficiency and scalability in software structure are needed to enforce and exemplify the context which is one of the technical issues. This issue was studied by Active project and treated by means of representation and reasoning approach. Shortly after, reservations with airlines companies, famous hotels, cinemas and most of the web services could be accessed by Siri (Winarsky, Mark and Kressel, n.d.). Furthermore, there was working on Siri to entitle it to read messages, set alarm, access to calendar, show maps, remind user friend’s birthday and tells stories and jokes (Hoy, 2018).
Figure 2. Setting the alarm task in Siri (European Commission, 2018).
3.2 Recent software Development
Unprecedentedly, Apple allowed applications developers in 2016 to access Siri which influenced amateurs to contribute to development as a third party alongside Apple (European Commission, 2018). In a bid to progress, Apple decided to launch HomePod device in December 2017 using the same virtual assistant Siri and associate to Home speaker Market with its competitors (Hoy, 2018). Hoy (2018, p85) also states that “Apple is also teaching Siri to recognize a user’s voice, but that feature had not yet been released at the time of this writing”. Finally, in April 2018, Apple announced the development of personalised tool “hey Siri” which can recognise users’ voices without pressing the Home Button. The aim of enabling this feature is to increase the privacy level “who is talking to Siri” which is a debatable issue in Hoy’s article (Siri Team, 2018).
It is clear from the previous section that Siri has made significant progress since 2011. The problems of natural language understanding, and voice recognition have been addressed and improvement was made. A number of updates have helped make Siri more developed and useful, for example, the inclusion of voice recognition feature may reduce the risk of hacking personal information. Conversely, Hoy (2018) mentions that there is a high possibility for an invader to attack a victim’s device by using an unheard command transform at ultrasonic frequencies. He also stated that Amazon’s assistant named Alexa faced the same security problems which demanded Amazon implement a system that recognise users’ voices. Since Alexa was added to the Amazon website interface, it has caused serious privacy issues. For example, items can be ordered by anyone utilising Amazon customers’ accounts via voice. As a result of this, there were several reported news pointing out that Alexa was listens to orders without distinguishing customers’ voices. One case on these reports, story which claimed happened to a Dallas girl aged 6 wanted to please her parents, she ordered £160 Kid-Kraft Sparkle Mansion and four pounds of sugar cookies using her parents’ account (Hoy, 2018).
Another negative aspect is that devices which have VAs listening to and recording every speech 24/7 which poses questions about where this information is stored. VAs use online clouds to store information, whereas VAs users speak to VAs to accomplish tasks or ask them to find information. All other conversations will be recorded such as health information, which pose a threat to personal privacy and could allow unlawful entities to use this information illegally or manipulate this personal information to imitate a user as described in Figure 3. Keeping devices constantly on may help hackers to expose and listen to voices and sounds in real time (Chung et al., n.d.)
Figure 3. compromised IVA-enabled devices (Chung et al., n.d.)
Furthermore, Apple offers a professional virtual assistant in its devices, but it charges its customers for hardware without charging for Siri. This leads to wide debate unanswered questions such as the funders who is paying for this massive platform and whether are we being observed or not (Stucke and Ezrachi, 2017). Therefore, all authors above support the idea that privacy is still a considerable issue in VAs and users’ privacy is in danger.
It can be concluded that considerable improvements have been made in Siri as a virtual personal assistant since 2011. Even so, there is an inability in terms of privacy, of keeping users completely protect from attack. Although Siri is a professional virtual personal assistant, the method used to store information in clouds increase the probability of hacking this information. Therefore, because of the growth of demand on VAs, it is crucial that further conserved information is needed. Due to the limitations of time and a lack of sources of Siri, this report made with sufficient research and required further research to support Siri’s limitation on privacy question.
- Chung, H. et al. (n.d.), ‘Alexa, Can I Trust You?’, Computer, 50, 9, pp. 100-104, Science Citation Index, EBSCOhost, [Accessed on: 19.07.2018].
- European Commission (2018), ‘The Rise of Virtual Personal Assistants’ in Digital Transformation Monitor, January 2018, (online), Accessible at: https://ec.europa.eu/growth/tools-databases/dem/monitor/sites/default/files/Virtual%20personal%20assistants_v1.pdf, [Accessed on: 19.07.2018].
- Hoy, M. (2018), ‘Alexa, Siri, Cortana, and More: An Introduction to Voice Assistants’, Medical Reference Services Quarterly, 37, 1, pp81-88, Scopus®, EBSCOhost, [Accessed on: 19.07.2018].
- Orner, R. (2016), ‘Siri (computer program)’, Salem Press Encyclopedia Of Science – Research Starters, EBSCOhost, [Accessed on: 18.07.2018].
- Pant, T. (n.d.), ‘Chapter 1: introduction to virtual assistant’ in:Building A Virtual Assistant For Raspberry Pi. The Practical Guide For Constructing A Voice-Controlled Virtual Assistant, pp1-8 [United States]: Apress, 2016., University of Liverpool Catalogue, EBSCOhost, [Accessed on: 17.07.2018].
- Purdy, EP. (2013), ‘Online privacy’, Salem Press Encyclopedia – Research Starters, EBSCOhost, [Accessed on: 18.07.2018].
- Reisinger, D. (2017), ‘Apple’s Latest iOS Update Looks Closer at AirPods and Siri’, Fortune.Com, p. 1, Business Source Complete, EBSCOhost, [Accessed on: 19.07.2018].
- Siri Team (2018), ‘Personalized Hey Siri’, Machine learning Journal,Vol. 1, Issue 9, April 2018 (online), Accessible at: https://machinelearning.apple.com/2018/04/16/personalized-hey-siri.html [Accessed on: 19.07.2018].
- Stucke, M., & Ezrachi, A. (2017), ‘How Digital Assistants Can Harm Our Economy, Privacy, and Democracy’, Berkeley Technology Law Journal, Issue 3, p. 1239, HeinOnline, EBSCOhost, [Accessed on: 19.07.2018].
- Wacks, R. (2015), ‘Chapter 1: Privacy in Peril’ in: Privacy: A Very Short Introduction (2015), pp1-33, Oxford: Oxford University Press, second edition, University of Liverpool Catalogue, EBSCOhost, [Accessed on: 19.07.2018].
- Winarsky, N., Mark, B. and Kressel, H. (n.d.), ‘The Development of Siri and the Siri Venture Creation Process’, Caltech Case HK, V.11.5 (online), Accessible at: http://www.mauldineconomics.com/images/uploads/overmyshoulder/Siri_Caltech_ Case_HK_-_Bill_Norman_v11_5.pdf, [Accessed on: 20.07.2018].