China robotics surge fuels security concerns

“On par with the cognitive level of a 10-year-old child.”

Wang Xingxing, founder of Chinese robotics manufacturer Unitree, used this phrase in a recent interview with Chinese media to describe the current state of robotics technology. Unitree’s humanoid robot G1 recently performed a group martial arts routine at China’s Lunar New Year gala. Wang’s comment suggests that the robot, capable of executing complex movements and martial arts sequences, now possesses cognitive abilities comparable to a fourth-grade student and is likely to advance even faster. Observing the rapid evolution of Chinese robotics over the past year, I have felt both awe at its progress and growing unease.

I first encountered a Chinese-made humanoid robot exactly one year ago. At Unitree’s headquarters in Hangzhou, Zhejiang province, the G1 waved to visitors and extended its hand for a handshake. While it had sufficient balance to avoid falling, its gait was awkward, and most movements depended on a human operator using a controller. At the time, a robot dog repeatedly performing backflips in place left a stronger impression than the humanoid machine.

The situation was similar during a half-marathon held in Beijing in April last year. Some robots could not even take a single step from the starting line. Assistants holding controllers often had to wait as the machines moved forward at a sluggish pace. Many of the robots resembled children aged three to five who require constant guidance and encouragement.

The transformation of China’s robotics industry soon became evident. By August, robots running on a track inside Beijing’s speedskating arena displayed movements much closer to those of humans. Assistants struggled to keep pace with machines running continuously at 12 to 13 kilometers per hour. Tian Gong, which finished second in a 1,500-meter race, ran autonomously without human control. In robot soccer matches, machines independently made decisions to shoot and pass, executing key plays on their own.

Last month’s Spring Festival gala delivered an even greater shock. Multiple robots performed in coordination with 8-year-old martial arts students on a narrow stage. Their reenactment of drunken boxing movements, including collapsing and rising again in fluid sequence, showcased tangible technological progress. A robot that seemed comparable to a 3- to 5-year-old child only a year ago now appeared to have matured more than twice over in that short period.

Another striking aspect of the gala was the robots’ use of swarm control technology. The machines ran at high speed while shifting formations with precision. Even when intermingled with the young martial artists during choreographed demonstrations, they showed virtually no margin of error. Drawing on preprogrammed algorithms combined with real-time data from sensors, the robots moved autonomously.

Chinese humanoid robots have displayed their capabilities across a variety of settings, including marathons, combat competitions, and Olympic-related events. Watching them move in disciplined formations and perform martial arts routines raises the unsettling thought that their next stage could one day be the battlefield.

Swarm control technology is considered a critical element of modern warfare. In particular, systems coordinating offensive unmanned aerial vehicles, commonly known as drones, are among the most advanced military technologies in global competition. The impact of drones on the battlefield has already been demonstrated in Ukraine and the Gaza Strip.

Although deploying humanoid robots in combat may still be a distant prospect, their large-scale introduction into industrial sites has already begun. The recent declaration by Hyundai Motor’s labor union that it would oppose the use of humanoid robots in factories reflects this trend.

Given that robots appear to have advanced the equivalent of five to seven human years within a single year, the pace of social transformation driven by robotics is likely to accelerate further. This development calls for deeper and faster reflection on how to safely manage robots equipped with artificial intelligence and how humans and machines can coexist sustainably.