This article examines the use of thermal imaging in human–robot interactions and its implications for privacy. Thermal imaging is used in robotics research to conceal the identity of individuals by masking identifiable information, such as facial details. The assumption is that by removing clear facial images, individuals become unidentifiable, thus protecting their privacy. However, the definition of personal data does not require information to be linked to a visual observation of a person. Removing the facial image of a person does not exclude the information from being personal data. However, thermal imaging can reveal personal information that is hidden from view. The article argues that privacy in robotics research is often limited to informational privacy and neglects other dimensions of privacy. It emphasises the need for a broader understanding of privacy in the context of robotics, considering the physical embodiment and data processing capabilities of robots. The conclusion of the article suggests that thermal imaging should not be automatically assumed to be more privacy-preserving than regular RGB cameras. It emphasises the importance of understanding personal data and the specific purposes of robot functions when choosing between different sensors. The article calls for a comprehensive approach to privacy in human–robot interaction, taking into account the broader impact of invasive technologies.