누구에겐 ‘눈과 귀’, 누구에겐 ‘감시 카메라’…AI 안경 피해 방지책 절실

A recent study has revealed that nine out of ten people in South Korea are concerned about the risk of being unknowingly filmed by AI glasses. While there is significant optimism about personal benefits such as real-time translation or navigation, the study indicates a dual approach in regulating usage for others. The report, "Artificial Intelligence Glasses: The Light and the Shadow - Innovation Broadly, Risk Precisely," was released by the Korea Research Foundation on the 15th.

The survey of 600 general citizens found that respondents felt more risk (average score of 4.20) than expected benefit (average score of 4.02) from the various functionalities of AI glasses. The greatest concern was "unauthorized photography or video recording" (89.5%) followed by "illegal filming or stalking" (87.2%) and "unauthorized recording of conversations" (86.5%).

Respondents showed a clear divide in their acceptance of recording and analyzing others, with only 26.8% willing to allow this compared to 56.2% willing to allow it for themselves. The acceptance of recording and analysis varied by location, with 48.2% of test-takers choosing a "principle of prohibition" compared to 26.8% of general users.

Additionally, 92.2% of respondents believed that cheating during exams using AI glasses would be a serious issue, but 82.7% believed that assistive devices for the disabled or medical purposes should be allowed with prior approval. The Korea Research Foundation suggests that because half of the respondents (48.7%) feel a "dual attitude," a conditional approach based on safety measures should be adopted for policy rather than a blanket ban.

They propose the introduction of "traffic light rules" that apply different regulations based on the sensitivity of the location (regular, sensitive, or high-risk) and the risk level of the functionality (basic, recording, and analytical functions). Kim Eui-hwan, director of the Korea Research Foundation's Artificial Intelligence Research Institute, emphasized the need to design precise rules based on where, by whom, and which functions are used before a practice becomes entrenched.