Page 164 - ISC PROCEEDINGS 21.4

H1: Perceived anthropomorphism has a positive impact on Privacy Disclosure Risk.
H2: Perceived anthropomorphism has a positive impact on Intrusiveness Risk.
Perceived autonomy refers to the extent to which a virtual streamer appears to act
independently, exhibiting self-governed behaviors without visible human intervention.
While autonomy enables seamless operation, Reactance Theory suggests that when
technology acts without user initiation, it threatens the user’s sense of freedom and
territorial privacy. This creates an "ethical paradox" in human-AI interaction. Aw et al.
(2024) argue that as AI agents become more autonomous, they are increasingly perceived
as intrusive to human identity. The ability to independently observe and process data
mirrors "surveillance technology" leading users to feel their personal space is being
invaded. Reinforced by Rohden et al. (2025), high autonomy is statistically linked to
heightened perceived intrusiveness. When a streamer proactively interacts without
explicit commands, users perceive this independent agency as a violation of privacy
boundaries, thereby increasing intrusiveness risk. Based on these arguments, we propose:
H3: Perceived autonomy has a positive effect on Intrusiveness Risk.
High perceived autonomy in virtual streamers directly amplifies privacy disclosure
risk due to inherent unpredictability. Ivanov et al. (2020) link this to "automation fears,"
where a lack of human control correlates with heightened risk assessments. This anxiety
is exacerbated by the "black box" nature of autonomous algorithms. Shin (2021) argues
that low explainability amplifies uncertainty, which Liu (2021) identifies as a primary
driver of perceived risk. Furthermore, Reactance Theory suggests that proactive,
unbidden interactions by AI can be perceived as intrusive violations of personal space
(Wang et al., 2025). Rohden et al. (2025) empirically confirmed that such "intrusive"
autonomy triggers a loss of behavioral control. Consequently, to protect their privacy
boundary against these unpredictable agents, users are likely to assess the risk of
disclosing information as significantly higher. Thus, the following hypothesis is proposed:
H4: Perceived autonomy has a positive effect on privacy disclosure risk.
Drawing on Equity Theory, consumers weigh the privacy costs they sacrifice
against the benefits received; when privacy risks outweigh benefits, users resort to
resistance as a mechanism to restore equity (Adams, 1965; Culnan & Armstrong, 1999).
Empirical studies confirm that high privacy risk triggers self-protective behaviors,
ranging from service avoidance to falsifying information (Herriger et al., 2025; Menard
& Bott, 2025). In AI contexts, the "watching-eye" effect and concerns over
unauthorized data merging heighten psychological anxiety, causing users to prioritize
risk avoidance over potential utilities (Hu & Min, 2023; Song et al., 2022).
Consequently, perceived privacy threats directly drive users to reject or discontinue
the use of technology to prevent potential losses (Choung et al., 2024; Mutimukwe et
al., 2020). Thus, the following hypothesis is proposed:
H5: Privacy disclosure risk has a positive impact on Resistance Intention.
Empirical evidence consistently demonstrates that heightened intrusiveness triggers
defensive resistance across various contexts (Aw et al., 2024; Sahli & Zhai, 2025; Shao &
Ho, 2025). In physical advertising, perceived intrusiveness was identified as the primary
predictor of direct rejection behaviors, such as the use of "No Junk Mail" stickers (Simon,
2016). Similarly, within social media, intrusiveness negatively affects attitudes toward
sponsored ads, thereby reducing user interaction intentions (Lin & Kim, 2016). In the
realm of smart devices, the "always-listening" capability of voice assistants fosters a sense
of intrusion, acting as a major barrier to adoption (Lucia-Palacios & Pérez-López, 2021; Pal

163