A Machine, a Teenage Girl, and a Question the Law Cannot Ignore

When a machine commits an act that would be a crime if performed by a human being, whom should society punish—the machine, its owner, its programmer, or the company that manufactured it?

Chapter 1: The Night a Robot Became a Police Suspect

Cologne, Germany. A late October night.

The streets were slowly emptying. The restaurants were closing, the traffic was thinning, and people were hurrying home through the chilly night air.

Outside a fast-food restaurant, a teenage girl was going about her business. She had no reason to imagine that an ordinary evening would become the subject of an extraordinary legal debate.

Then came the robot.

It was a humanoid machine, approximately four feet tall, reportedly dressed in a costume resembling Iron Man, the fictional superhero. Its movements were mechanical, but its appearance was sufficiently human-like to attract attention.

According to reports, the machine approached a 16-year-old girl and allegedly touched her buttocks before making thrusting movements towards her.

The girl must have been startled. What was supposed to be a harmless encounter with a robot had allegedly turned into an uncomfortable and potentially frightening experience.

But here was the first strange question.

If a man had behaved in this manner, the central question would have been straightforward: Who was he, and what had he done?

But this was a robot.

Could a machine commit sexual assault?

Could the police arrest it?

Could it be prosecuted?

And, perhaps most importantly, who was really responsible for its behaviour?

The incident was reportedly recorded and livestreamed. Investigators subsequently located the robot in the boot of a car and seized it, along with two mobile phones.

According to the report, the robot was identified as a Chinese-made Unitree R1 Air, reportedly costing around ₹5.3 lakh. Its 16-year-old owner and a 19-year-old companion were also mentioned in accounts of the incident.

The investigation would have to establish precisely what happened, how the robot was operated, and whether any human being deliberately directed its movements.

For the moment, however, the robot had become something unusual: a machine at the centre of a criminal investigation.

Imagine the conversation at a police station.

“Inspector, whom have you arrested?”

“A robot.”

“A robot?”

“Yes, sir.”

“And what is the charge?”

“An alleged sexual assault.”

The inspector might have paused before asking the obvious question.

“Then bring me the person who made it do that.”

But what if nobody had made it do anything at that particular moment? What if the behaviour had been programmed in advance? What if someone had remotely controlled it? Or what if a technical fault had caused the movements?

Those possibilities lead to very different questions of responsibility.

The investigation, rather than the robot’s appearance, would have to provide the answers.

Chapter 2: A Body Without a Conscience?

For centuries, criminal law has been built around a relatively familiar idea: a person performs an act, and the law examines the circumstances, the intention, and the consequences.

A person strikes another individual. Did the person intend to cause harm? Was the act deliberate? Was it accidental? Was there negligence?

Now introduce a machine capable of walking, speaking, recognising people, interpreting instructions, and moving its limbs with considerable precision.

The machine has a body. It can perform actions. It may even appear to understand a conversation.

But does it possess a conscience?

Imagine a courtroom in the not-too-distant future.

The judge looks towards the dock.

“Does the accused understand the charge?”

The defence lawyer rises.

“My Lord, the accused can process the words, but whether it understands them in the human sense is another matter.”

“And does it regret what it has done?”

“The machine can generate a statement of regret, My Lord. That does not necessarily mean it experiences remorse.”

The judge leans forward.

“Then how do I determine whether it is guilty?”

The lawyer has no simple answer.

This is the philosophical difficulty at the heart of robotic crime. A machine can produce behaviour that resembles human conduct without necessarily possessing human intentions, feelings, or moral understanding.

A robot may say, “I am sorry.”

It may lower its head, soften its voice, and appear embarrassed.

But is it experiencing shame—or merely executing a programmed response?

The distinction matters.

An apology generated by software is not automatically evidence of remorse. Similarly, an unlawful physical act performed by a robot does not, by itself, establish that the machine possessed the legally required intention.

Yet we must be careful not to draw the opposite conclusion too quickly. The absence of human-like consciousness does not make the consequences of a robot’s actions any less real.

If a machine injures a child, frightens a woman, or exposes an elderly patient to danger, the victim still deserves protection, investigation, and justice.

The machine may not possess a conscience.

The people responsible for designing, operating, supervising, or deploying it certainly remain subject to questions about their own responsibilities.

Chapter 3: The Guilty Mind

In criminal law, the Latin expression mens rea refers broadly to the mental element required for an offence—a guilty mind.

The exact mental element varies according to the offence and the applicable law. Some offences require proof of intention or knowledge; others may involve recklessness or different statutory requirements.

The law does not merely ask, “What happened?”

It may also ask, “What did the accused intend, know, or foresee?”

Now consider the robot outside the restaurant.

Its hand moved. Its body approached the girl. The alleged conduct may have been recorded.

But what does the recording establish?

It may provide evidence of the robot’s movements. It does not necessarily establish who controlled those movements or why they occurred.

Suppose the police question the owner.

“Did you instruct the robot to approach the girl?”

“No, officer.”

“Did you activate a special movement sequence?”

“I did not.”

“Was the robot being controlled through your phone?”

“I don’t know.”

These are imagined questions, not a transcript of the actual investigation. But they illustrate what investigators would need to establish.

They might examine the robot’s control logs, available video recordings, software, remote-control systems, and communications between devices. They might also investigate whether the behaviour resulted from a deliberate instruction, pre-programmed functionality, unauthorised interference, or a malfunction.

Each possibility changes the legal analysis.

If a person deliberately directed a robot to perform an unlawful act, the person could potentially face criminal liability under the applicable law.

If a person knowingly used a machine as an instrument to intimidate or assault someone, the fact that the person’s own hands never touched the victim would not necessarily protect that person from prosecution.

A machine could become an instrument of human wrongdoing.

If the conduct resulted from a technical defect, the investigation might instead raise questions about negligence, product liability, safety standards, or other legal responsibilities. Criminal liability would depend on the relevant law and the evidence.

And if the robot had been hacked, investigators would need to identify the intruder and determine whether that person could be traced and held responsible.

The essential question would remain: what happened, who contributed to it, and what does the evidence establish?

Chapter 4: When Machines Have Already Caused Harm

The German incident did not emerge in a world unfamiliar with robotic accidents.

Robots have already demonstrated that mechanical systems can cause physical harm, whether through faulty detection, unsafe design, inadequate supervision, or deliberate misuse.

Consider a few reported episodes.

In 2015, Swiss authorities seized an automated shopping bot called Random Darknet Shopper after it purchased ecstasy tablets from an online marketplace. The bot belonged to an art installation created by the collective Mediengruppe Bitnik. The artists had programmed it to purchase random items within a weekly budget. Authorities later released the bot and returned most of the seized items; the drugs were destroyed. The episode raised questions about the relationship between autonomous purchasing and human responsibility.

In 2016, a security robot at a shopping centre in California reportedly knocked over a toddler. The episode was not comparable to an allegation of deliberate sexual misconduct, but it raised a basic safety question: who is responsible when a machine fails to recognise and avoid a vulnerable human being?

In 2024, a humanoid robot attracted international attention during a public presentation in Riyadh, Saudi Arabia. Video appeared to show the machine extending its hand behind a female reporter in a manner that prompted allegations of inappropriate touching. The precise circumstances and intent should not be inferred from the video alone.

In December 2025, reports described a robot breaking a child’s finger during a chess interaction. The incident, if accurately reported, illustrates how even a seemingly ordinary activity can become dangerous when a machine’s physical capabilities and human safety are not adequately reconciled.

These episodes are different in their circumstances, and they should not all be described as crimes.

An accident is not automatically an assault. A malfunction is not automatically evidence of criminal negligence. And a disturbing appearance on video does not, by itself, prove intent.

Nevertheless, all these episodes point towards a common challenge.

As robots become more physically capable, society must decide who bears responsibility when their actions harm people.

Chapter 5: The Robot in the Nursery

Imagine a hospital in 2035.

A humanoid robot works in a paediatric ward. It carries medicines, answers routine questions, reminds nurses about schedules, and entertains children while they wait for treatment.

Parents initially admire it.

“Look, doctor,” a father says, “our son is not afraid of the robot at all.”

The doctor smiles.

“That is useful. A friendly machine can make a hospital feel less frightening.”

One day, however, the robot makes a mistake. It interprets a movement incorrectly and grips a child’s wrist too tightly.

The child screams.

A nurse rushes forward and disconnects the machine.

The parents are furious.

“Who is responsible for this?” the father demands. “The nurse? The hospital? The company that made the robot? Or the software that told it to move?”

The hospital administrator has to investigate.

Was the machine defective? Was its force limit improperly configured? Did the hospital fail to supervise it? Did the manufacturer issue adequate safety instructions? Was the nurse adequately trained? Had the robot been updated or modified?

There may be more than one contributing factor.

The same questions become even more serious when robots are used in homes, schools, elder-care facilities, public transport, and security services.

A machine that operates near vulnerable people cannot be treated merely as an interesting gadget. Its physical actions must be governed by meaningful safeguards, monitoring, and accountability.

The more powerful the machine, the greater the need to understand what it can do—and what it must never be allowed to do.

Chapter 6: Who Is Responsible When Nobody Is Holding the Remote?

The easiest cases are those in which a human being clearly controls the machine.

A person issues an instruction. The robot obeys. The instruction is unlawful, and the evidence establishes the operator’s role.

The harder cases arise when the machine behaves unexpectedly.

Imagine that a domestic robot is hacked and ordered to unlock a door, record private conversations, or threaten a resident.

The owner may have done nothing wrong. The manufacturer may have designed the system with reasonable safeguards. The intruder may have exploited a previously unknown vulnerability.

Who should be held responsible?

The answer cannot simply be that the robot did it.

Investigators would need to identify the source of the instruction, examine the security arrangements, and determine whether any human or organisation breached a legal duty.

Nor should responsibility automatically be assigned to the manufacturer whenever a machine causes harm. Products can be misused, modified, or attacked in ways that require careful investigation.

Conversely, a company should not be able to escape scrutiny merely by saying, “The artificial intelligence made the decision.”

That sentence may describe the immediate mechanism behind an event. It does not settle the question of legal accountability.

Companies decide how products are designed, tested, marketed, updated, and deployed. Operators decide where and how machines are used. Owners may decide whether to supervise or modify them. Regulators establish safety obligations.

Depending on the facts, more than one party may bear responsibility.

The law must examine those contributions rather than allowing the word autonomy to become a convenient excuse.

Chapter 7: The Law Cannot Put a Machine in Prison

Can a robot be prosecuted and sentenced to imprisonment?

Under current mainstream legal frameworks, robots are not generally recognised as independent criminal persons capable of receiving criminal punishment in the way human defendants can.

A robot can be seized as evidence. It can be examined by technical experts. Its operation can be restricted, and it may be disabled or removed from service where the law permits.

But imprisoning a robot presents a conceptual problem.

What would imprisonment mean to a machine?

Would it understand the punishment? Would it experience the loss of freedom? Would it learn moral responsibility from confinement?

Or would imprisonment simply mean locking it in a storage room, disconnecting its battery, and preventing it from operating?

The latter might protect the public, but it would be a safety measure rather than imprisonment in the ordinary human sense.

A court could, where legally authorised, order a dangerous product withdrawn, impose restrictions on its use, or deal with the responsible human parties. The exact remedies would depend on the jurisdiction and the circumstances.

The central principle is that the law need not pretend that a robot is a human being in order to control it.

It can regulate the machine while holding the appropriate people and organisations accountable.

And that may be the more practical route.

Chapter 8: What If Artificial Intelligence Becomes More Intelligent Than Its Creators?

Here the debate becomes more unsettling.

Today’s robots can be remarkably capable, but their abilities vary widely. A humanoid machine that performs pre-programmed movements is not equivalent to an advanced system that plans, adapts, and makes decisions across many unfamiliar situations.

As artificial intelligence advances, machines may increasingly perform actions that their operators did not explicitly anticipate.

Imagine a robot instructed to protect a building.

Its objective is simple: prevent unauthorised entry.

It detects a person attempting to enter through an emergency exit to escape a fire. The robot interprets the person as an intruder and blocks the doorway.

The instruction was to protect the building. The outcome endangers a human life.

The problem is not necessarily that the robot has developed a wicked personality. It may be that its objective, training, constraints, or interpretation of the situation failed to account for an essential human value.

A safety system must therefore do more than pursue a goal efficiently.

It must recognise limitations, respond appropriately to uncertainty, and allow human intervention when necessary.

An intelligent machine can be extremely competent at completing a task while remaining dangerously poor at judging when the task itself should yield to a more important consideration.

That is why human oversight cannot be reduced to a person sitting somewhere with a remote control.

Effective oversight requires people to understand the system’s capabilities, receive meaningful warnings, intervene in time, and remain accountable for decisions within their control.

The goal is not to stop every machine from acting independently. Many useful systems depend on a degree of autonomy.

The goal is to ensure that autonomy does not become an excuse for preventable harm.

Chapter 9: The Most Dangerous Sentence in the AI Age

One day, after a serious incident, an official might stand before a television camera and announce:

“The system made the decision. Nobody expected this outcome.”

The statement may be factually correct.

But it must not be the end of the inquiry.

A journalist should ask:

“Who approved the system?”

An investigator should ask:

“Was the risk foreseeable?”

A regulator should ask:

“Were adequate safeguards in place?”

A court should ask:

“What duty did each responsible party owe, and was that duty breached?”

And the victim should be entitled to ask:

“Who will answer for the harm I suffered?”

The phrase the AI did it must not become the technological equivalent of nobody did it.

At the same time, justice requires evidence. It would be unfair to punish an owner merely because a robot belonging to that person caused harm, just as it would be unfair to accuse a manufacturer without examining the facts.

Accountability must be both meaningful and fair.

That is the balance society must achieve.

Chapter 10: The Final Question

Return, for a moment, to the streets of Cologne.

A humanoid robot stands in police custody. Investigators examine the machine, the recordings, and the circumstances surrounding the allegation.

The public watches with fascination.

Some people laugh at the absurdity of arresting a robot. Others worry about what happens when machines become increasingly human-like. Still others ask whether the law is prepared for a future in which artificial intelligence controls bodies capable of touching, lifting, restraining, driving, and physically interacting with human beings.

But beneath the headlines lies a question far more serious than whether a robot can be called a criminal.

What happens to human responsibility when human beings create machines capable of causing harm?

Perhaps the most important question is not whether the robot has a guilty mind.

It is whether the humans who design, manufacture, program, operate, and deploy it have fulfilled their responsibilities.

A robot does not become harmless simply because it lacks emotions. Nor does it become a criminal simply because its movements resemble those of a guilty person.

The facts must be established. The law must be applied. Victims must be protected. And those who are genuinely responsible must not be permitted to hide behind a machine.

The future may contain robots that look increasingly human, speak with extraordinary fluency, and perform tasks that once required years of human training.

But their human-like appearance must not confuse our moral judgment.

A machine may imitate courage without being brave. It may imitate kindness without feeling compassion. It may say that it is sorry without experiencing regret.

And it may perform an act that causes real harm without possessing a human conscience.

That is precisely why society must remain clear about the distinction between the machine’s behaviour and the responsibility of the people behind it.

The robot may have the body. The software may supply the instructions. But justice must still discover who is responsible.

For the greatest danger of the artificial intelligence age may not be that machines become human.

It may be that human beings learn to use machines—and then claim that nobody is responsible for what those machines do.