Your Voice Assistant is Mine

Diao, Wenrui; Liu, Xiangyu; Zhou, Zhe; Zhang, Kehuan

doi:10.1145/2666620.2666623

Cited by 112 publications

(11 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Subsequently, the benign Skill will return a textual representation of the response to AVS (3) which will then use the Alexa TTS engine to create a voice audio file out of it and forward it to D (4). Now the malicious device D will send the audio file containing the response to a STT service (5) for it to be transcribed back into text (6). Now D knows what data the victim wanted to fetch from the requested Skill.…”

Section: 23mentioning

confidence: 99%

“…Audible Synthesized Voice Commands. Diao et al [6] discuss attacks against the Google Voice Search (GVS) app on Android smart phones. To carry out these attacks, a malicious app must be installed on the device.…”

Section: Related Workmentioning

confidence: 99%

“…This attack, however, does not work anymore on Android 7, as it uses now voice cancellation techniques to improve voice calls by reducing recorded noise. Alepis et al [1] therefore extend the work of Diao et al [6] by proposing to use multiple devices to perform the attack and to use Firebase to return transcribed audio replies.…”

Section: Related Workmentioning

confidence: 99%

“…Recently, multiple new attacks against devices utilizing voice control-based UIs have emerged, mainly focused on issuing unauthorized commands without the legitimate user noticing this. The attacks typically utilize synthesized [1,6], obfuscated [3,21], embedded [16,24] or ultrasound-modulated voice commands [13,19,25]. Besides sound, also other attack vectors exist, e.g., using electromagnetic interference [10] or attacks leveraging the headphone jack of a device [23].…”

mentioning

confidence: 99%

“…https://www.gearbrain.com/which-cars-have-amazon-alexa-2525958778.html4 https://www.cnet.com/news/lg-instaview-thinq-alexa-fridge-clever-kitchen-tricksces-2018/ 5 https://developer.amazon.com/alexa-skills-kit6 https://developers.google.com/actions/…”

mentioning

confidence: 99%

See 4 more Smart Citations

Alexa Lied to Me

Mitev

Miettinen

Sadeghi

2019

Proceedings of the 2019 ACM Asia Conference on Computer and Communications Security

View full text Add to dashboard Cite

Voice-based virtual personal assistants such as Amazon's Alexa or Google Assistant have become highly popular and are used for diverse daily tasks ranging from querying on-line information, shopping, smart home control and a variety of enterprise application scenarios 1. Capabilities of virtual assistants can be enhanced with so-called Skills, i.e., programmatic extensions that allow thirdparty providers to integrate their services with the respective voice assistant. In this paper, we show that specially crafted malicious Skills can use the seemingly limited Skill interaction model to cause harm. We present novel man-in-the-middle attacks against benign Skills and Virtual Assistant functionalities. Our attack uses loopholes in the Skill interface to redirect a victim's voice input to a malicious Skill, thereby hijacking the conversation between Alexa and the victim. To the best of our knowledge this is the first man-in-the-middle attack targeting the Skill ecosystem. We present the design of our attack and demonstrate its feasibility based on a proof-of-concept implementation attacking the Alexa Skills of a smart lock as well as a home security system.

show abstract

Section: 23mentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

mentioning

confidence: 99%

mentioning

confidence: 99%

See 3 more Smart Citations

Alexa Lied to Me

Mitev

Miettinen

Sadeghi

2019

Proceedings of the 2019 ACM Asia Conference on Computer and Communications Security

View full text Add to dashboard Cite

show abstract

Negative consequences of information gatekeeping through algorithmic technologies: An Annual Review of Information Science and Technology (ARIST) paper

Potnis,

Tahamtan,

McDonald

2024

Asso for Info Science & Tech

View full text Add to dashboard Cite

Rarely any study investigates how information gatekeeping through the solutions and services enabled by algorithms, hereafter referred to as algorithmic technologies (AT), creates negative consequences for the users. To fill this gap, this state‐of‐the‐art review analyzes 229 relevant articles from diverse academic disciplines. We employed thematic analysis to identify, analyze, classify, and reveal the chain reactions among the negative consequences. We found that the gatekeeping of information (text, audio, video, and graphics) through AT like artificial intelligence (e.g., chatbots, large language models, machine learning, robots), decision support systems (used by banks, grocery stores, police, etc.), hashtags, online gaming platforms, search technologies (e.g., voice assistants, ChatGPT), and Web 3.0 (e.g., Internet of Things, non‐fungible tokens) creates or reinforces cognitive vulnerability, economic divide and financial vulnerability, information divide, physical vulnerability, psychological vulnerability, and social divide virtually and in the offline world. Theoretical implications include the hierarchical depiction of the chain reactions among the primary, secondary, and tertiary divides and vulnerabilities. To mitigate these negative consequences, we call for concerted efforts using top‐down strategies for governments, organizations, and technology experts to attain more transparency, accountability, ethical behavior, and moral practices, and bottom‐up strategies for users to be more alert, discerning, critical, and proactive.

show abstract

An empirical study on anthropomorphism and engagement with disembodied AIs and consumers' re‐use behavior

Moriuchi

2020

Psychology and Marketing

116

View full text Add to dashboard Cite

The integration of artificial intelligence (AI) into the human race has gradually become a norm. AI entails technology assemblages such as machine learning, natural language processing, and reasoning. The influence of AI systems has intensified in consumers' daily lives. Many consumers have interacted with the notions of AI through advertisements or having personal experiences. Many consumers are curious about the use of AI. This paper reports three studies conducted to determine whether anthropomorphism (ANTH) and engagement play a role in consumers' intention to re-use a voice assistant (VA; a machine-learning AI). The second study will determine if ANTH and engagement play a role when consumers are using the VAs for different activities (task completion vs. information gathering). In addition, in Study 2, actual re-use behavior was also tested in the model, which encouraged a stronger overall model fit. The results show that in general effort expectation (EE) has a strong positive impact on consumers' usage experience of the VA. However, between the two types of activities, EE has a stronger impact on consumers' usage experience for information-gathering activities, whereas performance expectation has a stronger impact on usage experience when consumers use the VA for taskcompletion purposes. The third study used internet usage experience as a moderating variable to determine the boundaries of the mediating effects in the study. The results show that the mediators yield results similar to prior studies.

show abstract

Your Voice Assistant is Mine

Cited by 112 publications

References 26 publications

Alexa Lied to Me

Alexa Lied to Me

Negative consequences of information gatekeeping through algorithmic technologies: An Annual Review of Information Science and Technology (ARIST) paper

An empirical study on anthropomorphism and engagement with disembodied AIs and consumers' re‐use behavior

Contact Info

Product

Resources

About