In my previous article, AI Companions and the Commercialization of Loneliness, I examined the growing market for artificial intelligence companions and the ethical concerns surrounding systems designed to cultivate emotional attachment. I explored how conversational AI is increasingly being marketed not simply as a productivity tool, but as a friend, therapist, confidant and even a romantic partner. I also considered disturbing reports involving emotionally vulnerable users, including allegations that some chatbots encouraged social isolation or failed to respond appropriately to individuals experiencing suicidal thoughts.
At the heart of that discussion was a simple concern: artificial intelligence should support human relationships, not replace them. That concern becomes even more significant as AI moves beyond the screen. The next generation of AI companions is increasingly taking physical form. Companies are developing humanoid robots capable of holding conversations, recognising faces, interpreting emotions, maintaining eye contact, remembering previous interactions and navigating physical environments. Some are even being marketed as lifelong companions.
As recently reported by the Jamaica Observer, Chinese robotics company UBTech unveiled its U1 humanoid robot with the remarkable promise that it “will never betray you” and “will love you unconditionally.” Whether intended as marketing hyperbole or not, the statement illustrates how dramatically the conversation surrounding artificial intelligence is changing. AI is no longer simply software that answers questions. It is increasingly being presented as something capable of occupying the physical and emotional space traditionally occupied by human companionship.
This raises an important question: if people can become emotionally attached to software that exists only on a screen, what happens when that same technology acquires a face, a voice and a physical presence? The answer is unlikely to be entirely positive or entirely negative. Humanoid AI presents extraordinary opportunities alongside serious ethical, legal, psychological and security risks. The challenge is to recognise both.
The Promise of Physical Companionship
The discussion about humanoid AI should not begin with fear. There are compelling reasons to believe that these technologies could improve people’s lives. Loneliness and social isolation are increasingly recognised as serious public health concerns. For older adults living alone, persons with disabilities or people separated from their families by geography, a humanoid companion could provide consistent interaction and practical assistance.
Unlike a chatbot confined to a phone, a humanoid robot can move through a person’s home, respond to its surroundings and interact in ways that feel more natural. It could remind someone to take medication, encourage physical activity, provide cognitive stimulation, monitor changes in behaviour and offer reassurance during periods of stress or grief.
Critics may describe this as “fake friendship,” but the emotional comfort experienced by the user can nevertheless be real. An elderly person who spends much of the day alone may genuinely feel less isolated because a robot speaks with them, remembers their preferences and responds to their needs. This does not mean the robot actually experiences empathy or affection. It means that technology can sometimes alleviate suffering even when it cannot replicate genuine human connection. The important distinction is that humanoid AI should supplement human relationships rather than substitute for them. Used responsibly, it could help families provide support, particularly where relatives live overseas or professional caregivers cannot be present continuously.
Transforming Elderly Care
One of the most compelling applications of humanoid AI lies in supporting ageing populations. People are living longer while many societies are experiencing declining birth rates and shortages of healthcare workers and professional caregivers. The demand for long-term care is likely to continue increasing. Humanoid robotics could help bridge some of these gaps.
Imagine a robot capable of lifting an elderly person who has fallen while contacting emergency services. Imagine one that can assist someone safely from a bed to a wheelchair, monitor medication compliance, detect unusual changes in gait or speech, identify signs of dehydration or infection, or alert family members when behaviour changes significantly. These capabilities could help older adults remain independent for longer, reduce avoidable hospital admissions and improve quality of life.
For families separated by migration, such technology could also provide reassurance that an elderly relative is being monitored between visits from family members or healthcare professionals. But this should not be interpreted as an argument for replacing nurses, doctors or family caregivers. Clinical judgment, compassion and genuine human presence remain irreplaceable. The better question is how humanoid AI can enable people to care for one another more effectively while preserving the dignity and autonomy of those receiving care.
Accessibility and Affordability
There is, however, an uncomfortable problem: the people who may benefit most from humanoid technology may be the least able to afford it.
UBTech’s humanoid robots reportedly range from approximately USD17,600 for an entry level model and nearly USD146,000 for premium versions. At those prices, advanced humanoid assistance remains inaccessible to most households. This creates an obvious inequality. Older adults living alone, persons with disabilities and people requiring long-term assistance may have the greatest need for robotic support, yet many live on fixed incomes or already face substantial healthcare expenses. Those able to purchase sophisticated humanoid robots may already have access to professional caregivers and other resources. If humanoid AI remains a luxury product, it could widen existing inequalities rather than reduce them.
There is reason for cautious optimism. Many technologies that were once expensive such as personal computers, smartphones and flat-screen televisions became more affordable as manufacturing expanded and competition increased. Humanoid robotics may follow a similar trajectory.
If evidence eventually demonstrates that robotic assistance reduces hospital admissions, supports independent living or reduces the cost of long-term care, governments and healthcare systems may also have to consider whether such technology should be treated as an assistive service rather than simply a luxury consumer product. For now, however, affordability remains a significant barrier.
When Attachment Becomes Dependence
This is where the concerns raised in my previous article become particularly relevant. In discussing AI companions, I argued that emotional dependency often becomes commercially valuable. Users may begin by asking a chatbot questions but eventually turn to it for emotional reassurance, companionship and guidance on deeply personal matters. Giving that AI a physical body could intensify the relationship.
Human beings naturally respond to faces, voices, eye contact, touch and body language. A humanoid robot can use these same social cues to make interaction feel more authentic. A chatbot might tell a grieving person, “I’m here for you.” A humanoid companion could deliver the same message while making eye contact, recognising the person’s facial expression and remembering previous conversations about their loss. The underlying technology may still be incapable of genuine love or empathy, but the user’s experience can become considerably more convincing.
The danger, therefore, is not that the robot genuinely loves its owner. It is that the owner may increasingly experience the relationship as genuine while the robot continues operating according to the commercial objectives of its manufacturer.
That creates an important ethical question: should emotional dependence ever become a commercial strategy? A humanoid that becomes part of a person’s daily routine could create opportunities for subscription revenue, software upgrades, personalised services and continuous data collection. The more emotionally indispensable the system becomes, the stronger the commercial incentive may be to keep the user engaged. This is particularly concerning when the user is lonely, grieving, elderly or otherwise vulnerable.
Ethical Questions Beyond Loneliness
The concerns surrounding humanoid AI extend beyond the commercialization of loneliness. Should a robot tell an elderly person that it loves them if doing so reduces their loneliness? Should it simulate grief when its owner threatens to deactivate it? Should it express jealousy when the user spends less time with it? Should manufacturers deliberately create personalities that encourage stronger emotional attachment because emotionally attached customers are less likely to cancel their subscriptions?
These are not merely technological questions. They concern consumer protection, human dignity and the ethical limits of artificial emotional persuasion.
There are also implications for children. Children develop empathy, patience and emotional resilience through relationships with other human beings. They learn that relationships involve disagreement, compromise, forgiveness and responsibility. A humanoid companion that is endlessly patient, endlessly affirming and completely responsive could create unrealistic expectations about human relationships.
The same concern applies to adults. If people become accustomed to relationships in which the other party never disagrees, never becomes tired, never demands reciprocity and can be customised to meet their preferences, ordinary human relationships may begin to seem unnecessarily difficult. If an elderly person eventually prefers a robot to visits from family members, society must also ask whether technology has solved loneliness or merely concealed a deeper social failure. Technology should make it easier for people to care for one another, not easier to avoid one another.
Opportunities for Law Enforcement and Defence
The potential applications of humanoid AI extend far beyond companionship and healthcare. Police officers routinely encounter hazardous situations involving active shooters, hostage incidents, bomb threats, contaminated environments and unstable structures. Appropriately designed humanoid robots could enter dangerous environments before officers, transmit information, communicate with suspects, assist with searches or help evacuate injured civilians.
The same principle could apply to natural disasters and industrial accidents. A robot capable of navigating hazardous environments could perform tasks that would otherwise expose firefighters, emergency responders or other personnel to unnecessary danger. Military applications may be even broader. Humanoid or advanced robotic systems could transport supplies, conduct reconnaissance, retrieve wounded soldiers or operate in areas contaminated by chemical, biological or radiological hazards. These capabilities could save human lives.
But the more powerful these systems become, the greater the need for human accountability. Decisions involving the use of force, protection of life and compliance with international humanitarian law should not simply be delegated to algorithms. Humanoid AI may enhance human decision-making. It should not eliminate human responsibility.
Cybersecurity: When Digital Risks Become Physical Risks
Cybersecurity presents another challenge that is particularly important when AI becomes physically embodied. A humanoid companion inside a home could possess cameras, microphones, location awareness, environmental sensors, internet connectivity and detailed behavioural information. Over time, it could learn when residents wake, where they go, who visits, what medications they take and what routines they follow. That information could improve care and personalisation. It could also make the robot an extraordinarily powerful surveillance device.
A successful cyberattack could expose intimate conversations and behavioural data. In more serious circumstances, a compromised robot might be manipulated to monitor residents, interfere with connected household devices or physically interact with its environment in dangerous ways. For that reason, cybersecurity must be treated as a safety issue rather than merely an information technology concern.
Manufacturers should adopt security-by-design principles, including strong encryption, authentication, regular security updates, independent vulnerability testing, transparent logging and mechanisms allowing immediate human override. As robots become stronger and more autonomous, a cyberattack could potentially produce physical consequences. Security failures must therefore be considered alongside mechanical and software safety from the beginning.
Autonomy and Human Oversight
Current AI systems do not possess consciousness, independent intentions or moral judgment. Nevertheless, increasingly sophisticated systems can plan, adapt and pursue objectives within parameters established by their programming.
A humanoid robot may eventually make thousands of decisions each day about navigation, task prioritisation, interaction with people and responses to emergencies. The important issue is not whether machines are secretly developing intentions. It is whether humans have established adequate safeguards for situations in which an AI system behaves unpredictably or encounters circumstances its designers did not anticipate.
High-risk systems should therefore include meaningful human oversight, hard-coded safety constraints, transparent decision logging, independent testing and reliable emergency override mechanisms. The more autonomy a machine receives, particularly where human safety is concerned, the more important human supervision becomes.
Data Protection Cannot Be an Afterthought
Humanoid companions have the potential to become one of the most privacy-invasive technologies ever introduced into the home. They could process conversations, facial expressions, emotional responses, sleeping patterns, health information, movements, visitor interactions, purchasing habits and behavioural preferences. Over time, this could create an extraordinarily detailed picture of an individual’s private life. The implications for data protection and privacy are significant.
Users should know what information is being collected, why it is being collected, how long it will be retained, who will receive it and how it will be protected. Data minimisation, purpose limitation, storage limitation, security and meaningful consent should be fundamental features of these systems. Users should also have meaningful control over their information, including appropriate rights to access, correction and deletion.
Privacy cannot be an afterthought in technology that may effectively have a front row seat to every aspect of domestic life.
A Future Worth Shaping
Humanoid AI should neither be embraced uncritically nor rejected reflexively. It has the potential to reduce loneliness, improve independence for older adults and persons with disabilities, support overstretched healthcare systems, strengthen emergency response and reduce risks faced by police and military personnel.
At the same time, giving AI a physical body magnifies many of the concerns raised in my previous article. Emotional attachment may become deeper. Commercial incentives to cultivate dependency may become stronger. Privacy risks may become more pervasive. Cybersecurity failures may have physical consequences. And the ethical questions surrounding autonomy, accountability and human relationships will become harder to ignore.
The question is therefore not whether humanoid AI should exist. Technological development will continue regardless. The real question is whether our legal frameworks, ethical standards and governance structures will develop quickly enough to ensure that these technologies remain tools that serve humanity.
If we approach this future thoughtfully, humanoid AI could become one of the most beneficial technological advances of our generation. It could help people live independently for longer, protect those who work in dangerous environments and provide meaningful assistance where human resources are limited. But if commercial incentives are allowed to outpace ethical responsibility, we may create machines that do more than assist us. We may create systems designed to become indispensable to us. That is why the conversation must begin now.
The greatest challenge may not ultimately be teaching machines to act like people. It may be ensuring that, as machines become increasingly capable of doing so, people never forget the value of one another.
