Certain humanoid robots can be compromised via voice commands, a demonstration reveals.

Researchers are making progress in understanding the complexities of robotics and cybersecurity. Yicaiglobal reported (via Interesting Engineering) that specialists from the Chinese cybersecurity research group Darknavy recently demonstrated a method of compromising commercial humanoid robots, highlighting important implications for the future. At GEEKcon in Shanghai, the researchers showcased a commercially available Unitree robot, revealing it could be manipulated using voice commands. Even more interestingly, they showed how it could potentially be used to affect other nearby units.

The robot operates with an internal AI agent, and by identifying a software flaw, the researchers were able to gain control of the robot while it was connected to a network. They demonstrated how the robot could communicate wirelessly to spread the issue to another nearby robot, which was not even connected to the network at the time. Remarkably, this spread occurred in just a matter of minutes. Additionally, the researchers were able to command the robot to interact with a mannequin on stage, further illustrating their findings.

The implications of this research are significant and thought-provoking. Historically, cybersecurity concerns have primarily focused on the theft of money or information. As humanoid robots become more common in our homes, particularly for elder care and other supportive roles, new challenges emerge. This could open avenues for individuals with malicious intent to cause harm, raising concerns regarding safety and security.

For now, it is reassuring to know that these demonstrations were conducted in a controlled environment, with a focus on identifying and resolving potential vulnerabilities rather than amplifying them. Taking proactive steps to understand and address these challenges contributes to a safer future for all.