Unitree G1 EDU Robot Vulnerabilities Allow Root RCE

Security researcher Olivier Laflamme has disclosed two independent root remote code execution (RCE) vulnerabilities affecting the Unitree G1 EDU humanoid robot. These flaws enable an attacker to gain complete control over the robot's systems. The first vulnerability, tracked as CVE-2026-76639, allows for a network-adjacent path to achieve root access. This exploit chain involves the chat_go and bashrunner components. The second vulnerability, identified as CVE-2026-76640, presents a more critical threat as it can be exploited through Bluetooth Low Energy (BLE). This BLE-based exploit allows an attacker to achieve root privileges on the robot's Locomotion PC, which is the primary processing unit for the robot's movement and coordination. The Unitree G1 EDU is an educational model of the Unitree G1 humanoid robot, designed for research and development purposes, often utilized in academic institutions and robotics labs to explore advanced locomotion and AI capabilities. The disclosure of these vulnerabilities raises significant concerns regarding the security of advanced robotic systems, particularly those that may operate in environments with potential network or wireless exposure. Root RCE is a severe type of security exploit that grants an attacker the highest level of access to a system, allowing them to execute any command, modify system files, install malware, or steal sensitive data. The fact that one of these vulnerabilities is accessible via Bluetooth Low Energy is particularly concerning, as BLE is a common wireless communication protocol used in many IoT devices and can have a relatively short range, making it accessible to attackers in close proximity. This proximity requirement, however, does not diminish the severity of the exploit, as it bypasses typical network-based security measures. Unitree Robotics, the manufacturer of the G1 EDU, is expected to address these vulnerabilities through firmware updates or security patches. The implications of such vulnerabilities extend beyond the immediate risk to the specific robot models, highlighting a broader need for robust security practices in the development and deployment of humanoid robots and other advanced AI-powered systems. As robots become more integrated into various sectors, including research, industry, and potentially public spaces, ensuring their security against malicious actors is paramount to maintaining trust and preventing potential harm. The detailed technical analysis of these flaws, provided by Laflamme, will be crucial for Unitree and the wider cybersecurity community in developing effective countermeasures. The specific nature of the chat_go and bashrunner components suggests potential weaknesses in how the robot handles inter-process communication or command execution, while the BLE vulnerability points to potential issues in the wireless stack's authentication or authorization mechanisms. Further investigation into the exact implementation details of these components will be necessary to fully understand the attack vectors and develop comprehensive mitigation strategies.