Two Unitree G1 EDU Humanoid Robot Flaws Enable Root RCE, One Starts Over Bluetooth
What happened
Two separate vulnerabilities enabling root remote code execution have been found in the Unitree G1 EDU humanoid robot. Security researcher Olivier Laflamme disclosed these issues, now tracked as CVE-2026-76639 and CVE-2026-76640. One of the vulnerabilities leverages a path through the robot’s Bluetooth Low Energy interface to gain root access on its Locomotion PC. The other involves a network-adjacent vector linked to components called chat_go and bashrunner.
The risk
These flaws allow attackers to take full control of a critical robot system remotely, including through Bluetooth. The Bluetooth route is especially concerning because it does not require network access, lowering the barrier for an attacker positioned nearby. Root access means the attacker can modify software, disable protections, or manipulate the robot’s behavior—raising serious safety, privacy, and operational risks.
Why it matters
Unitree G1 EDU robots are used in education and research, so these vulnerabilities expose not just enterprise or consumer deployments but also academic and development environments. The ability to remotely hijack such robots can halt projects, increase costs through incident response, and introduce liability issues around physical safety. Operators must patch quickly and reassess use of Bluetooth interfaces in their security models.
Who should pay attention
Anyone deploying Unitree G1 EDU robots, including educators, researchers, and developers using them for experimentation or automation, must be alert. Security teams in organizations with physical robot fleets need to include these devices in vulnerability management and physical security reviews. Bluetooth-enabled robotics platforms gain a new profile as a risk vector.
What to watch next
Follow updates from Unitree Robotics and security advisors for patches and mitigation strategies. Expect increased scrutiny on Bluetooth attack surfaces in robotics. Researchers and operators should watch for related disclosures across similar robotics platforms, as these flaws could inspire similar exploits elsewhere in the market.
AI Quick Briefs Editorial Desk