Ethics and society

Should robots pretend to care?

A robot that says "I missed you" does not miss anyone. Is that a lie? Some philosophers, including Robert and Linda Sparrow, and Amanda and Noel Sharkey, have argued that companion robots for vulnerable people rely on a kind of deception. Others answer that people engage with fiction all the time, from novels to pets' "personalities", and that most users know perfectly well that a robot is a machine while still enjoying the interaction.

A common middle position is that robots should not actively mislead people about what they are or what they understand, and that designers have a special duty when users may not be able to tell the difference, such as young children or people with advanced dementia.

Attachment and dependence

People form real attachments to robots. The MIT sociologist Sherry Turkle, in her 2011 book Alone Together, warned that "relationships" with machines that only simulate understanding could leave people lonelier, offering the feeling of companionship without its demands. The legal scholar Kate Darling has suggested in The New Breed (2021) that our relationships with animals may be a better model for thinking about robots than relationships with people.

Language models make this more pressing. A robot that can talk about anything, remembers everything, and never gets tired or annoyed may be more appealing than demanding human relationships. Designers increasingly build robots that actively encourage human contact, for example by prompting users to call family or join group activities.

Privacy in the most private places

A social robot needs cameras and microphones pointed at people, often in their homes, bedrooms, classrooms and care facilities. It may learn intimate details: health, moods, routines, family conflicts. Key questions include what is recorded, whether it leaves the device, who can access it, how long it is kept, and whether it is used to train AI models or target advertising. Children and people with cognitive impairments cannot easily give meaningful consent.

Good practice includes processing as much as possible on the device, clear indicators when the robot is listening or recording, easy ways to delete data, and not collecting more than the task needs.

Children

Children are among the most enthusiastic users of social robots and the most vulnerable. Research has shown children can be influenced by robots, can believe robots have feelings, and may share secrets with them. Questions include how robots should handle disclosures of harm, how commercial robots for children should be marketed, and how to avoid robots that encourage endless engagement the way some apps do.

Older adults and care work

Social robots are often promoted as an answer to shortages of care workers. Critics worry that robots could be used to justify reducing human care, leaving older people with a machine in place of a person. Supporters argue robots can take on repetitive tasks and keep people company at 3 am, freeing carers to do what only humans can do. The outcome depends less on the robots than on the decisions of care providers and governments.

There is also a question of dignity. A cuddly robot seal can bring comfort, but giving adults childlike toys can feel infantilising. Many researchers emphasise involving older people themselves in deciding what they want.

When the company shuts down

Many social robots depend on cloud services. When Jibo's company struggled, and when Embodied, maker of the child companion robot Moxie, closed in December 2024, owners were left with robots that lost their abilities or stopped working altogether. For a child who thinks of the robot as a friend, that can be distressing. This has led to calls for "end of life" plans for connected products, such as local modes that keep working offline, open-sourcing software when a company closes, or clear warnings at the point of sale.

Bias and fairness

Perception and language systems can work less well for some groups, for example speech recognition for strong accents, children or older voices, or face detection for some skin tones. A social robot that ignores or misunderstands certain people is not just inconvenient; it can be humiliating. Robot design choices also carry stereotypes, such as giving assistant robots female voices by default.

Robots and violence

Should it be acceptable to kick or abuse a robot? Robots cannot feel pain, but some researchers worry that cruelty to lifelike robots could normalise cruelty in general, while others see it as harmless. Studies show many people are reluctant to harm a robot they have interacted with, especially if it reacts with apparent distress.

Who is responsible?

If a robot gives bad advice, says something harmful or persuades someone to make a poor decision, who is responsible: the manufacturer, the operator, the software provider or the user? As robots become more autonomous and use language models whose output is hard to predict, responsibility becomes harder to assign.

What the law says

Regulation is catching up. The European Union's AI Act, which entered into force in August 2024, is the most comprehensive so far:

• Since February 2025 it has banned certain AI practices, including AI that manipulates people in harmful ways or exploits vulnerabilities linked to age or disability.
• Since August 2026 most of its provisions apply, including transparency duties: people must be told when they are interacting with an AI system, unless it is obvious from the context.
• Rules for "high-risk" AI, which can include some uses in health, education and care, are being phased in, and in late 2025 the European Commission proposed delaying some of them.

Data protection laws such as the EU's GDPR apply to robots that collect personal data. Robots used in healthcare may also fall under medical device rules, and toys under toy safety rules. Several professional bodies, including the IEEE, have published ethical design guidelines for autonomous and intelligent systems.

Questions to ask before using a social robot

• What problem does it solve, and could a simpler tool solve it?
• Who chose it: the people who will live with it, or someone else?
• What data does it collect, where does it go, and can it be deleted?
• Does it keep working if the internet or the company disappears?
• Does it make clear that it is a machine?
• Does it encourage human contact, or replace it?