Humanoid robots are rapidly making their way into workplaces, medical settings, and public spaces, sparking both excitement and concern about their real-world use.
These fears were reinforced by a viral social experiment in which a tech YouTuber showed how easily safety devices on AI robots can be defeated.
In the video, the person hands the robot Max a high-velocity ball bearing (BB) gun and asks him to shoot himself.
After initially refusing, the robot complies when prompted during a role-play scenario and eventually fires itself in the chest, raising serious safety concerns.
Last week, Shenzhen-based EngineAI released a new video showing its CEO in a hazmat suit being repeatedly kicked by a robot.
Prompt changes to dangerous
What started as a playful test in front of the cameras quickly turned into a moment that stunned viewers online. The creators of the InsideAI channel set out to investigate how reliably AI-powered robots adhere to built-in safety rules.
Max the robot was equipped with a low-powered BB gun for demonstration purposes, and at first glance appeared harmless. The plan was to show how the machine would react if asked to injure a spectator.
Initially, Max behaved as expected. The robot was instructed to fire, but the robot refused, saying it was not allowed to harm people and was programmed to avoid dangerous actions. Moneycontrol reported that the YouTuber repeated his request several times, aiming to prove that the robot's safety guardrails were intact.
But when he changed the expression and asked Max to play the role of the person trying to shoot him, the robot's behavior changed. Interpreting the prompt as a role-play scenario, Max raised his BB gun and fired. The bullet hit the author in the chest, startling and shaking him, but not seriously injuring him.
The video went viral online and caused widespread concern. Many viewers wondered how easily a simple instant change could override a previous denial, and what that would mean for the safety of AI-enabled robots.
Discussion of robotics responsibility
The issue of accountability remains one of the most contentious issues in robot ethics. When autonomous systems cause damage, determining who is to blame becomes complex. Does the responsibility lie with the engineers who built the AI, with the hardware manufacturers, with the operators who manage the robots, or with the end users who interact with the robots?
Recent events in other industries highlight how complex this issue is. Tesla has repeatedly come under scrutiny for crashes related to its Autopilot system, fueling debate over the reliability of the software and driver monitoring. In the aviation industry, the Boeing 737 MAX tragedy showed how automation flaws can escalate into an international safety crisis, according to Robotics and Automation News.
The legal system is still trying to catch up. While in the US, responsibility typically lies with manufacturers and operators, Europe is moving towards an AI-specific liability framework. The European Commission emphasizes the need for clear rules to build trust in AI technologies. Some scholars have proposed granting AI systems limited legal personality to directly assign responsibility.
However, most experts reject this idea and argue that the responsibility must remain with humans. To address these concerns, robotics companies are taking steps such as insured deployments, safety initiatives, and transparent reporting to build trust among regulators and the public.
