Deceived by Immersion: A Systematic Analysis of Deceptive Design in Extended Reality

We began with the development of the search protocol, where we defined the search query and databases. To ensure the inclusiveness of our search, we targeted four bibliographic databases: Scopus,⁶ the ACM Guide to Computing Literature,⁷ IEEE Xplore,⁸ and JSTOR.⁹ Both the ACM Guide to Computing Literature and IEEE Xplore offer a strong focus on technology- and engineering-related publications, Scopus has multi-discipline publications, and JSTOR provides access to humanities and social sciences journals. Given that XR is a cluster of technological innovations, and deceptive design takes advantage of human weaknesses, we decided to include all four databases in our initial search. The search results from all four databases together offered a good balance in depth and breadth for our review.

In addition, we defined a set of search terms using keywords that frequently appeared in deceptive design definitions (as described in Section 2.2), as well as terms frequently used to describe XR technology. We conducted multiple iterations of searches with these keywords to test and refine the combinations of search terms and to ensure that the research outcomes fit our scope. Table 1 presents the final search query for individual databases.

Our search was based on “Abstracts” instead of “Full Text,” given that Scopus and JSTOR do not allow “Full Text” searches. We intended to keep our search query as consistent as possible. Thus, the “Abstract” search was applied to all four databases. Further, our search terms excluded nouns such as “manipulation” and “deception” because these terms were frequently used to describe manipulating factors in controlled experiments or skills in surgical procedures (e.g., laparoscopic manipulation skill), and to describe brain activities in neuroscience (e.g., brain activity in deception and truth telling). Moreover, acronyms of XR technologies and terms such as “trick,” “steer,” “mislead,” and “subvert preferences” were excluded, since these were frequently used in other disciplines to represent irrelevant topics. Our final search queries include terms that describe XR technology, including “Virtual Reality,” “Augmented Reality,” “Mixed Reality,” and “Extended Reality,” and terms that describe deceptive design, including “dark pattern,” “deceptive design,” “deceive,” “manipulative,” and “abusive.” We include the detailed search query for each database in Table 1.

Last, we gathered a total of 187 publications including 181 articles from bibliographic databases and six workshop papers. We downloaded the titles and abstracts of the 187 publications in a spreadsheet. We manually inspected each and excluded 31 duplicates with identical titles and DOIs. In the end, we arrived at \(n=156\) unique records for the next phase.

With three coders, we first screened the 156 publications’ titles and abstracts based on the following exclusion criteria. Then, each publication was screened by two coders. In case of disagreement, the disagreed publications (4 publications) were passed on to the third coder who acted as a tie-breaker. The exclusion criteria were as follows:

A total of 18 publications were retained after the screening process. We then conducted a full-text screening of each, following the same exclusion criteria. Two coders screened the publications individually, and the third coder acted as a tie-breaker. We excluded 5 publications that only mentioned XR deceptive design as an example in the abstract but did not closely discuss the details in the paper. We retained \(n=13\) records after the screening process. The final sample includes 7 records from the four bibliographic databases and 6 records from the CHI workshops. We then conducted a thematic synthesis on these 13 publications.

We aggregated four ways XR technologies improve user experience from 10 publications: (1) product previewing, (2) experience previewing, (3) memory remembering, and (4) realistic experience and sensations. However, these publications simultaneously raise concerns about the potential for user deception under the guise of improving user experience. Among these publications, there was a strong agreement that XR technologies can enhance user experience with immersive, interactive, and photorealistic virtual environments. For instance, in their study on XR advertising, Mhaidli and Schaub [51] mentioned the IKEA Place AR-based app¹¹ that lets customers digitally sample furniture in their living area to better comprehend its size, design, and function. Similarly, before booking a hotel room, clients can join a virtual tour to “sample” the experience [51]. These preview features, while seemingly convenient, could present an ideal version of reality, potentially leading to uninformed purchases and disappointment when encountering the actual product [51]. In addition, the hypothesis of Bonnail et al. [5] and the analysis of Schlembach and Clewer [67] of a VR documentary suggest that XR has the potential to enhance personal and historical memory reminiscence as it provides “realistic” reconstruction of past events in 3D, and Krauß [36] described a VR documentary (“Meeting You”¹²) that enables people to spend time with deceased friends and families “as if they were still alive” (p. 1). This also raised ethical concerns regarding the potential for distinguishing between real and false memories, and the potential for impersonation and psychological manipulations [67]. Several papers noted that XR technology can produce stereoscopic images, music, and convincing haptic feedback to make people feel psychologically present in a virtual environment [11, 64, 84]. Beyond presence, Ramirez et al. [64] further noted in their theoretical argument paper that carefully designed VR simulations can generate “virtually real experiences,” where users engage with the virtual experiences as if they were real. This, achieved through a combination of high context-realism (i.e., the degree to which the rules, settings, and appearance of a simulated world respond to users as if they were real) and perspectival fidelity (i.e., a simulation that contributes to generating a user’s viewpoint), may further blur the line between virtual and real and enables XR to be used for manipulation purposes [64].

Nine publications discussed how XR features increase user manipulation. These publications consistently argue that XR-based deceptive designs are more convincing and manipulative compared to non-XR deceptive designs [5, 20, 38, 51]. From our literature analysis, we identified six such XR features that can contribute to user manipulation.

T2.1: The Illusion of Objectivity in XR Experience. To maximize the level of immersion, many XR games and applications promise a first-person experience that allows users to “see” the virtual world from the perspective of a certain individual, animal, or entity. An example is the 6x9 VR game that simulates a prisoner’s first-experience [28]. However, Ramirez et al. [64] argued that XR can never produce an objective first-person experience because the simulations always suffer from semantic variance [68] and structural intersectionality [10]. The potential for creators to inject their own perspectives into the content raises ethical concerns around XR simulations that claim to deliver a first-person experience [64].

T2.2: Obscuring Reality, Disguising Risks. As mentioned in Section 2, XR has the ability to obstruct and substitute the physical objects with a virtual environment. In a multi-user scenario, Buck and McDonnell [8] discussed the potential for one user to mislead another regarding their physical surroundings. This concern was echoed by Lebeck et al. [38] through AR user interviews, which illustrated users’ concern that XR allows hiding of dangerous physical items (e.g., a gun) behind virtual objects, causing harm to people using XR. However, Mhaidli and Schaub [51] noted that in XR, users cannot just look away like on a 2D screen, especially when they are unaware of the threats.

T2.3: Data Fuels Privacy Risks and Manipulation. Several publications emphasized that XR technologies process lots of data to support its user experience, such as data describing user body and eye movements [8, 38, 39], gestures [39, 51], physical appearance [8, 51], behavioral patterns (e.g., gait, mannerisms) [8, 38, 51], and physical surroundings [8, 38, 51]. The user experiments conducted by Miller et al. [54] and Pfeuffer et al. [62] revealed that XR users can be deanonymized based on their behavioral patterns, a type of data that commonly is considered non-private. These data can be a valuable resource for predicting user preferences and vulnerabilities, which enables the development of tailored manipulation strategies [51].

T2.4: Unawared Access, Unawared Data Use. Beyond single-user environments, three studies focused on multi-user interaction and application data misuse. For instance, the user studies conducted by Lebeck et al. [38] documented cases where users modify each others’ choice architectures by manipulating (e.g., hiding) each other’s virtual objects. Moreover, Lebeck et al. [38] revealed several problematic multi-user behaviors, such as when users place inappropriate virtual objects in the shared space or onto others’ avatars or damage others’ virtual objects without permission. Regarding information misuse, Lebeck et al. [38] identified users’ concerns regarding accidental disclosure of private information without understanding that others can see it and bystanders’ concerns of being unwilling participants in others’ XR experiences. Buck and McDonnell [8] further warned that these data might be abused by the XR applications or users, causing unexpected privacy risks and social consequences. This theme differs from the previous one in that the user was neither actively involved nor aware of the data collection process.

T2.5: Insecurity Worsens User Manipulation and Privacy Concerns. Several papers highlighted that current XR systems lack security controls that usually exist on mobile, PC, or web environments, leaving users vulnerable to manipulation and data privacy threats. For instance, due to the absence of the same-origin policy in WebVR [16], Lee et al. [39] prototyped a defense system to prevent attackers to inject ads or alter content on VR webpages [39]. Moreover, the experiment of Su et al. [73] demonstrated cases where attackers could steal user information by side channeling the VR hardware, and Buck and McDonnell [8] expressed concerns about the migration of Internet scams into the metaverse.

T2.6: Persistent Exposure to Manipulation. Two papers further expressed the concerns regarding user manipulation from a constant immersive experience [12, 51]. As companies are now developing daily-use XR equipment (i.e., Mojo smart lens¹³), Mhaidli and Schaub [51] hypothesized that future users will wear AR devices constantly throughout their daily life. As a result, Cummings and Shore [12] and Mhaidli and Schaub [51] warned that users will experience constant data collection and user surveillance, and their exposure to targeted advertising and XR-based manipulation will also become persistent.

We argue that Theme 2 revealed XR’s potential to enable deceptive design across all six attributes [48, 49]. The illusion of objectivity can be employed to induce false beliefs through users’ perceived objectivity of their experiences. XR’s capacity to obscure reality may facilitate “information hiding.” User data from XR sensors enables “covert” and “asymmetric” deceptive design that exploits users’ cognitive biases and preferences. The insecurity of XR systems allows for all types of deceptive design that exploits XR design elements and users’ false beliefs in the authenticity of XR content.

Eight publications discussed psychological manipulation through XR technologies [5, 51, 67], which we detail in this section.

T3.1: False Memory Implantation. A central feature of XR technology is immersion, where users are fully engaged and absorbed within a virtual environment [86]. Bonnail et al. [5] and Schlembach and Clewer [67] emphasized the memory manipulation risks posed by the immersive experience. As detailed in Theme 1 (Section 4.2.1), XR is capable of supporting memory reminiscing with “realistic” multi-dimensional reconstructions. However, Bonnail et al. [5] described the theory behind human memory flaw and argue that distorted XR reconstructions can lead to false memories. Specifically, humans suffer from source confusion and thus may confuse where their memories originated [69]. For example, a conversational rumor may be mistaken for TV news [5, 69]. Based on a UK VR advertising company’s “memory relive” service (Momento¹⁴), Bonnail et al. [5] hypothesized a scenario of a VR wedding reconstruction, where the VR service provider altered the champagne and wedding dress brands and customers falsely believed that they enjoyed this champagne during their wedding and bought from the brand again.

T3.2: Artificial Prosthetic Memory and Empathy-Based Manipulation. Drawing on the literature on memory, Schlembach and Clewer [67] mentioned that VR can be used to create prosthetic memories of historical events. Prosthetic memory is a public cultural memory people acquire from movies and television [37]. It fosters a personal connection to others’ historical experiences, and can create people’s empathy, social responsibility, and political alliances across race, class, and gender [37]. For example, people who had no personal connection to the 9/11 victims can became vicariously traumatized after watching broadcast media and films that exposed them to victims’ sufferings [32]. Humanities and social sciences literature believed that VR offers a route to empathy because it allows people to “walk in another’s shoes” and “see the world through another’s eyes” [66]. Through the examination of two VR films, Schlembach and Clewer [67] warned against “empathy generation” through VR technology as people can be “made to ‘feel’ something they will be changed by it and so will their behaviour.” They illustrated this argument through the example of a VR game (6x9) that simulates the psychological effects of extreme isolation resulting from solitary confinement, including distorted vision, hallucinations, floating, and screaming [28, 67]. This immersive experience evoked public empathy toward prisoners and led to the reform of legal and political decisions (i.e., in 2016, the Obama administration introduced federal changes to curb solitary confinement) [78].

In parallel, drawing from literature on the ethics of choice-architecture, Ramirez et al. [64] argued that the VR-based empathy-enhancing simulations are always unethical because it can never objectively provide actual experience of another, as we discussed in Theme 2 in Section 4.2.2. Thus, it may mislead users “about the nature of their virtual experiences,” and manipulate them into “changing their thoughts, beliefs, and behaviours,” consequentially influencing their “future judgments about their own values, policies, and so on” [64, pp. 530–532]. In addition, taking inspiration from a U.S. military recruitment game (i.e., America’s Army¹⁵), Mhaidli and Schaub [51] hypothesized that such games in VR may be more powerful in driving players’ inclination toward enlisting. They argued that immersive and realistic experiences of VR could create a sense of “living” in the military, and the enjoyment of the game might “bias” players to view joining the armed forces positively [51].

T3.3: Hyper-Personalized User Manipulation. Personalization is an e-commerce strategy that employs data to enhance experience of people with similar traits through tailored content, services, and features [51, 83]. Hyper-personalization refers to creating highly customized content tailored to an individual [51, 85]. As detailed in Theme 2 (Section 4.2.2), XR’s data collection capabilities enable inference of user states and facilitate manipulation that takes advantage of user data. Drawing inspiration from IKEA’s AR furniture preview application,¹⁶ Mhaidli and Schaub [51] discussed how AR-driven hyper-personalization can manipulate people’s purchase decisions by making “the photorealistic rendering of the furniture seem brighter and more colorful than real life.” Two papers echoed this discussion and illustrated other possible cases of data-based user manipulation in XR [8, 51]. For example, psychological user manipulation through idealized interaction partners (meeting their preferences) based on biometric data [8], emotional manipulation in user investment decisions using holographic recreations of their trusted people based on shared images [51], and physical navigate the user to a restaurant based on physiological signal of hunger [51].

In summary, we classify this theme as “deceptive,” “covert,” and “information hiding” [48, 49], as it takes advantage of users’ false belief in the authenticity of XR simulations, hides information about the ads’ inclusion, and manipulates them through mechanisms such as the photorealistic graphics and sensory feedback that are not directly apparent. We also argue that this theme partially but not entirely aligns with the pre-established “asymmetric” attribute of Mathur et al. [49] as the XR simulation subtly influences user decisions without any interface friction.

Several publications discussed user manipulation through seamless reality distortion and the idea that XR equipment will be worn daily [5, 38, 51]. Drawing upon the deceptive ads in the 2020 U.S. presidential campaign,¹⁷ Mhaidli and Schaub [51] suggested a potential scenario where future politicians could exploit daily-worn XR devices, concealing evidence of poverty and economic downturn in a city with photorealistic graphics. Bonnail et al. [5] echoed this scenario by proposing a potential use of VR technologies for marketing, where people’s uploaded photos could be modified to incorporate brand logos and create false memories of experiencing a product. The AR user stody of Lebeck et al. [38] further emphasized users’ safety concerns in AR-assisted driving, where deceptive representations of reality, such as hidden pedestrians or a misrepresented neighboring vehicle, could cause harm to both the driver and people nearby. The distorted reality can lead to psychological addiction because people “identify with the bodies that they virtually inhabit, even when those bodies are very different from their own” [64, p. 532].

We classify this theme as “asymmetric,” “covert,” and “deceptive” because it increases people’s exposure to ads, utilizes subtle influence mechanisms in XR that are unapparent to users, and capitalizes on people’s false beliefs by modifying the representation of reality [48, 49].

While our Theme 3 (Section 4.2.1) discussed XR’s ability to offer immersive virtual experiences, four publications highlighted its manipulation on user perceptions [8, 39, 51, 73].

T5.1: Extreme Realism Blurs the Boundary between Virtual and Reality. Mhaidli and Schaub [51] suggested that a skillfully crafted commercial with highly realistic visuals might fool viewers into thinking it was real. AR users may not realize that other people’s shirts include digital brand logos. Although current XR systems cannot yet generate such realistic simulations, Mhaidli and Schaub [51] and Buck and McDonnell [8] anticipated that it will be possible in the near future. In addition, Mhaidli and Schaub [51] and Krauß [36] noted XR’s product previewing feature to sway user buying decisions (e.g., brighter and more colorful furniture). The virtually real experience in XR makes the preview seem authentic, resulting in disappointment when the product is of lower quality [51].

T5.2: Perception Hacking Manipulates User Input. Using XR technologies’ visual, auditory, and haptic feedback to engage with the virtual world, the lab experiment of Su et al. [73] and the user study of Lee et al. [39] both showed that perception hacking in XR can effectively deceive people. A well-known non-XR example of perception hacking is clickjacking, which misleads users into accidental pressing of ads on websites (i.e., disguised ads) [6, 31]. Our analysis synthesized three types of XR perception hacking: (1) cursorjacking [39, 73], (2) blind spot tracking [39], and (3) auxiliary display abuse [39].

As a variation of clickjacking on the web, cursorjacking in XR manipulates visual cues of object movements (e.g., hand movements) to differ from proprioceptive cues (e.g., physical sensing of hand movements) [39, 73]. Research indicated that visual perception is valued more than other senses, even if the difference is small [3, 35]. The user study of Lee et al. [39] showed that an angular offset of visual cues from the user’s hands can redirect hand movement in VR and deceive them into interacting with undesired virtual objects. Taking inspiration from ad impression frauds on websites and mobile apps [41, 72], in a 360-degree immersive XR environment, Lee et al. [39] demonstrated blind spot tracking that uses gaze analysis to display ads beyond people’s line of sight. Similar to blind spot tracking, Lee et al. [39] also demonstrated auxiliary display abuse that occurs when participants are immersed in XR and therefore fail to notice ads on a secondary monitor (e.g., a PC monitor) [39].

In short, we classify this theme as “asymmetric,” “covert,” “deceptive,” and “information hiding” [48, 49], as it increases users’ exposure to specific brand logos, trick users into interacting with undesired virtual objects through subtle influence mechanisms, and capitalizes on people’s false beliefs of perceived body movements and hides the existence of ads outside of their line of sight.

Several publications expressed a concern that deceptive designs in XR may lead to greater and harder-to-defend privacy loss compared to those in non-XR [5, 12, 36]. For example, to enjoy the XR memory reconstruction service, customers must share personal images and video recordings, as “the more video provided, the more accurate the construction” [5, p. 3]. Based on diverse user data, XR produces superior experiences compared to non-XR systems, making it more challenging for users to prioritize privacy [12, 74]. In addition, XR applications can deceive users into disclosing private information to digital humans or those who resemble their loved ones [8, 36, 51]. Further, Cummings and Shore [12] proposed a possibility where privacy concerns may hinder users from reporting encountered deceptive design, potentially providing a “shield” for such issues in XR.

There was also a strong agreement that deceptive design in XR can lead to security risks [8, 38, 73]. For example, Su et al. [73] and Buck and McDonnell [8] proposed that perception hacking (Theme 5 in Section 4.3.3) can deceive people into accidental granting of privileges. Lee et al. [39] further noted that the insecurity of XR browsers leave space for ad fraud, and Su et al. [73] mentioned that attackers can exploit these security weaknesses to steal sensitive user data or distort XR content.

Another theme frequently mentioned by the publications is that misleading design induces artificial feelings that change people’s judgment of values, views, beliefs, morals, and politics [20, 51, 67]. For instance, the analysis by Schlembach and Clewer [67] of the 6x9 VR simulation game [28] elicited that empathy-based manipulations are effective in political and social control. Mhaidli and Schaub [51] mentioned another example of concealing poverty and economic decline to show a false bright economic environment for political campaigns. Users’ false beliefs in XR’s objectivity lead to “made” empathetic responses to manipulative simulations, which in turn affects their future thoughts, attitudes, and emotions [64]. Furthermore, the idealized digital humans, as suggested by Buck and McDonnell [8] and Franklin [20], can potentially be used for “political duress” or manipulating users into engaging in risky or unlawful behaviors, such as online gambling.

Several publications mentioned that people’s manipulated attitude, belief, thought, and perception can influence their behaviors [8, 20, 38, 39, 64]. For instance, the distorted preview of products and experiences [51], the use of a hyper-personalized ad spokesman [8, 51], and the integration of ads in XR memory reconstructions can impact people’s buying decisions [5]. In addition, Franklin [20] gave an example that players’ behaviors in video gameplay can “spill over” into their behaviors outside of game (i.e.,Virtual Spillover [63]). Thus, they argued that the same effect may occur among XR users, especially when XR exposure becomes constant [20].

Our analysis found eight publications suggesting solutions to protect people from XR deceptive design [5, 38, 57, 64]. For example, Franklin [20] and Nijholt et al. [57] recommended using speech and text analysis, physiological sensors, and intelligent software to detect and remove behavior- and preference-changing methods. To address the multi-user problems, Lebeck et al. [38] suggested creating methods and policies to prevent XR devices from overlaying virtual on others’ content and allowing users control over view and edit permissions for personal virtual objects and space.

T8.1: Preventing False Memories and Empathy-Based Manipulation. In the cases of memory manipulation, Bonnail et al. [5] proposed basing XR memory reconstruction on user consent, with service providers openly disclosing potential modifications, and identifying updated items in scenes to prevent source confusion and false recollections. Since empathy-based manipulation plays a crucial role in XR deceptive design, Ramirez et al. [64] proposed changing simulation goals from empathy-provoking first-person experiences to bystander perspectives. This reduces manipulation and engages users sympathetically instead of empathetically [64].

T8.2: Security and Privacy Recommendations. Five publications provided privacy and security recommendations for developing future XR technologies [8, 12, 39]. For example, Cummings and Shore [12] indicated that XR users needs to trust the information they consume and the services that use their data to freely and safely share information. Buck and McDonnell [8] argued that power should be given to the user, and they should be able to opt-out of data collection at any moment. In addition, Su et al. [73] advocated for enhancing security in XR systems, and Lee et al. [39] called for creating XR-specific security policies. Cummings and Shore [12] specifically mentioned that children should be protected with special laws and standards, and three publications agreed that XR data collection and misuse for malicious purposes should be prohibited [8, 12, 51].

T8.3: Call for Academic Research Efforts to Better Understand the Problems. Three studies called for research about problems in XR deceptive design [8, 20, 51]. For instance, Buck and McDonnell [8] suggested researching on psychological user manipulation in XR ads and the effects of physical properties of avatars. Buck and McDonnell [8] and Franklin [20] recommended researching how digital interaction translates into people’s daily lives. Mhaidli and Schaub [51] called for research on what preference changes are ethical or manipulative based on user benefits, and proposed studying privacy policies of XR advertisers and developers to understand XR data risks.

T8.4: Improve User Literacy to Identify and Resist Manipulation. Three publications emphasized educating users to raise awareness and resilience toward deceptive design and its risks [8, 12, 51]. Specifically, Mhaidli and Schaub [51] proposed improving users’ literacy of recognizing photorealistic XR ads with distorted previews and hyper-personalized avatars. They argued that enhancing users’ literacy in XR ads could improve their resistance to manipulation [51]. Buck and McDonnell [8] and Cummings and Shore [12] called for user education on XR dataflow and privacy implications. According to Mhaidli and Schaub [51], developing effective educational interventions will require academic and industrial research.

The findings of RQ1 (see Section 4.1) showed that existing research primarily relied on deceptive design definitions that emphasize designer intentions (e.g., [8, 57, 81]). This trend shows a need for XR deceptive design research using an updated definition and focusing on those that resulted from unintentional design decisions. Moreover, to effectively mitigate harmful deceptive design in XR, it is critical to understand to what extent the design mechanisms are considered benevolent (e.g., nudging) and to what extent they are considered deceptive (e.g., tricking). Gray et al. [22] and Gray et al. [24] suggested that users’ perceptions of manipulation can help identify borderline practices that may not be strictly problematic or illegal. During our literature analysis, we found papers that tried to differentiate “nudging” and ‘manipulation” based on user interest and freedom of choice (e.g., [51, 64]). Therefore, we suggest that future deceptive design research should include a comprehensive user-centric interaction or perception analysis. This may include exploring questions such as these: How do users perceive XR deceptive design for different purposes? and Which designs are considered manipulative and need to not be built, and which are acceptable? Without a clear understanding of users’ perspectives, developing effective and user-friendly countermeasures to deception will be challenging.

Our analysis uncovered deceptive designs from non-XR environments that are now extended to XR. We also observed new forms of deceptive designs that emerged from XR features. However—given that XR is still in its infancy and XR devices have yet to become mainstream—most of these studies were based on hypothetical scenarios of potential future problems. More problematic issues are likely to emerge with the advancement of XR technology. Therefore, we call for further research on new forms of XR deceptive designs and their impacts on users. That way, countermeasures can be developed to address these challenges before they can seriously harm people.

Among the publications we reviewed in our study, there was a noteworthy agreement that XR data enables deceptive design [8, 38, 54]. The extensive collection of user data through XR devices creates opportunities for precise user profiling, which can be used for hyper-personalization and manipulation [8, 51, 73]. To better comprehend and prevent this issue, we suggest that future research should start by asking what data types can be collected by XR sensors. The investigation by the XR Safety Initiative lab’s 1st XR data classification roundtable conference [87] and Dick [14] in 2021 have laid the groundwork for examining how VR and AR use data. However, several questions still need to be answered. These include questions such as the following: What user information can be inferred from the XR sensor data (especially from MR sensors)? How can these data be exploited in XR deceptive design? How does the use of these data impact user autonomy, safety, and privacy? What controls do users want to have over these data? Addressing these questions and developing appropriate protections for XR technology users require the continued efforts of researchers, developers, and designers.

Last, our findings in T8 (Section 4.4.3) emphasized the need for user education on deceptive design in XR [8, 12, 51]. Research has shown that users, despite being able to recognize deceptive design, often have a resigned attitude due to their dependence on the services [45]. Conversely, Bongard-Blanchy et al. [4] found that users’ vague awareness of the entailed concrete harms on themselves made them failed to realize the necessity of taking self-protective actions. Therefore, we argue that it is crucial to design effective user educational interventions. Such interventions should instruct people that XR may not always reflect reality. Users should be taught about cues that differentiate fictional simulations from realistic representations. Moreover, it is essential to inform people about XR data tracking, potential privacy issues, and XR mechanisms that could manipulate them to share their data. Most importantly, users should understand the possible consequences and risks of XR deceptive design. They should also receive training to help them detect scams, digital (i.e., fake) humans, and bots that pretend to be someone they trust. Future research is also needed to explore optimal educational content design for improving user understanding, measuring user learning outcomes, and assessing user resilience against deceptive design in XR.