Page 169 - ISC PROCEEDINGS 21.4

H5. Meanwhile, PDR has a stronger positive effect on RI (β = 0.356, t = 7.680, p < 0.001),
thereby supporting H6. Detailed results are presented in Table 7.
Table 7. Path coefficients and hypothesis testing results

Original Sample Standard T statistics P values
sample (O) mean (M) deviation (|O/STDEV|)
(STDEV)

PAN -> IR 0.263 0.263 0.069 3.792 0.000

PAN -> PDR 0.194 0.193 0.064 3.055 0.002

PAU -> IR 0.223 0.226 0.072 3.098 0.002

PAU -> PDR 0.117 0.120 0.072 1.629 0.103

IR -> RI 0.235 0.238 0.052 4.501 0.000

PDR -> RI 0.356 0.357 0.046 7.680 0.000

Source: Authors’ own work (2026)
3.4. Discussion
This study provides empirical evidence that perceived anthropomorphism has a
direct and positive effect on both intrusiveness risk and privacy disclosure risk in the
context of virtual streamers. The findings indicate that when virtual streamers exhibit a
high level of human-likeness, viewers no longer perceive them merely as neutral
technological tools but rather as social agents capable of participating in and influencing
their personal experiences. This finding is consistent with the Uncanny Valley Theory,
which suggests that artificial entities with a high degree of human resemblance but
lacking complete naturalness may evoke feelings of unease among observers. In
livestream environments, cues such as emotional expressions, flexible communication,
and personalized responses can blur the boundary between humans and machines. When
this boundary becomes unclear, users may perceive technological systems as having the
capacity to intervene more deeply in their personal space and experiences, thereby
increasing perceptions of intrusion risk. This result also supports the argument of Blut et
al. (2021) that highly anthropomorphic technologies may weaken users’ perceptions of
personal boundaries. Similarly, consistent with Thomaz et al. (2020), the human-like
characteristics of technology may reduce users’ perceived control during interactions. In
addition, within the context of commercial livestreaming, anthropomorphism not only
generates a sense of discomfort but also translates into specific risk evaluations related to
privacy. Consistent with Shao (2024), when the human-like features of virtual streamers
become more salient, users may begin to question the underlying motives behind such
interactions. In this situation, anthropomorphism may be interpreted as a form of social
façade that conceals functions related to data collection and processing, thereby
increasing users’ perceptions that personal information disclosure may occur in an
environment that is not entirely transparent.
Second, the results confirm that perceived autonomy has a direct and positive
impact on intrusiveness risk. When a virtual streamer is perceived as capable of operating
and making decisions independently, users may feel that their control over the
interaction process is reduced, thereby increasing their perception of potential intrusion

168