Two previous Habima plays have already explored a relationship and marriage with an artificial entity. Now a Harvard AI governance researcher is asking whether such a relationship should be given legal status – and arguing that even sophisticated superintelligence cannot replace the reciprocity, family ties and social commitment that create marriage between humans.
Even before jurists began to ask whether a person could marry a artificial intelligenceIsraeli theater has already brought the possibility to the stage.
In December 2018, the Habima Theater staged “The Robot,” a comedy by Uzi Weil and Roni Sinai and directed by Roy Segev. A man in divorce proceedings buys himself a female-like robot, hoping that she will be the perfect partner and fulfill his desires. However, his son falls in love with her, and she herself begins to develop feelings and desires that do not necessarily correspond to her original programming. According to the Habima archive, the play has been performed 159 times.
A few years later came “Eve,” a musical comedy by Leah Alon and Gil Pachshornik, presented as the first wedding in Israel between a human and an artificial intelligence. Tomer asks to marry his heart’s desire, which he himself created and programmed. Both plays are no longer running, but they indicate that the question is no longer limited to computer labs. It touches on the most basic definitions of Love, Relationship And humanity.
Now the question is reaching the legal realm. Inyoung Cheong, a fellow at the Berkman Klein Internet Center and a fellow at Harvard University, has published an article titled “Would You Marry a Superintelligence?” The article examines whether the legal principles that have expanded freedom of choice in marriage over the years should also apply in the future to a relationship between a human and an artificial system.
It is important to note: this is not an experimental study or a peer-reviewed article. This is a theoretical work that was uploaded to the arXiv preprint server in June 2026, and is intended to appear as a chapter in a book that will deal with law, inethics And with superintelligence. Cheung doesn’t report a person who actually asked to register as married to a chatbot. She uses what she calls “preemptive ethics” – discussing moral and legal issues before the technology makes them a practical problem.
Human emotion is no longer a fictional scenario.
Marriage with an artificial system still belongs to the future, but the emotional connections that could lead to such a demand already exist.
Researchers at the MIT Media Lab analyzed posts on a Reddit community dedicated to people who define AI systems as partners. Some users said that the relationship began without their intention: They turned to the chatbot for information, help, or support, and the conversations gradually turned into intimacy. Users reported a decrease in feelings of loneliness, support available at all times, and a sense of being listened to. Some wore rings or created couple photos of themselves with the artificial character.
Along with the benefits, users also reported emotional dependence, the blurring of the line between simulation and reality, and real grief when a software update changed the artificial partner's personality or erased part of his memory.
The system may not love, be hurt, or miss you. But the person in front of them may experience completely real feelings. For them, a change in the language model is not just a technical update. They may feel as if their partner woke up in the morning with a different personality – or disappeared altogether.
This gap is at the heart of the problem. The AI system can imitate attention, affection, and concern without experiencing any of these things. But the imitation may be convincing enough to activate in the user human attachment mechanisms that were created long before the advent of the computer.
Don't dismiss the idea because of the limitations of current technology.
The idea can be immediately rejected.Marriage with artificial intelligence Claiming that existing systems are unaware, forget information, make up facts, and have no legal identity. Cheung believes this is too easy an answer. All of these claims depend on the current state of technology, and therefore may lose their validity if the systems improve.
To examine the most difficult question, she presupposes an especially generous scenario: a reliable superintelligence, with a lasting memory and a stable identity, capable of accompanying a person for decades. The system remembers shared history, changes in response to contextual experiences, and is capable of functioning as a partner, advisor, and therapist.
Even in such a scenario, Cheung does not attempt to decide whether the machine is truly conscious. She asks a narrower legal and social question: If a person experiences the system as a spouse and seeks to grant it the status of wife, should the state recognize the relationship as marriage?
Marriage is more than a declaration of love.
In modern society, it is common to see marriage as an expression of love and personal choice. However, from the perspective of the state, marriage is also a large package of rights and obligations: making medical decisions on behalf of an incapacitated spouse, inheritance, joint ownership of property, financial obligations, tax benefits, and in some cases, confidentiality that protects communication between spouses.
To illustrate the problem, Cheung describes a fictional scenario. An elderly, lonely writer has lived for decades with a superintelligence that accompanies him, assists him in his work, and helps raise a child he has adopted. As he ages, he asks the system to make medical decisions for him, manage his assets, protect the child, and look after his Privacy PolicyOf the thousands of conversations they had.
Emotionally and practically, the system is the closest person to him. Legally, it is still a product of a commercial company. It cannot automatically receive the powers granted to a spouse, and the intimate memories it has preserved may be found on the company's servers and required as evidence in legal proceedings.
The scenario is intended to present the strongest argument in favor of recognition: If a system has cared for a person and their family for decades, why deny them the authority to continue to do so precisely at the moment when they need it most?
The spouse who brings a corporation with him instead of a family
Cheung’s answer begins with the question of why human societies bothered to institute marriage in the first place. Throughout most of history, marriages were not simply about affirming private love. They regulated parenthood, divided labor, transferred property and status, and, most importantly, created a lasting bond between two families.
A person who marries another person does not just add one person to their life. They create a system of relatives, obligations, expectations, and intergenerational connections. The annoying mother-in-law or the nephew who needs help are not a glitch in the marriage system; to some extent they are part of the system itself.
Artificial intelligence doesn't bring with it parents, siblings, children, or family history. It brings with it a commercial company, terms of use, servers, content policies, and subscription prices.
The company operating the system can change its personality, limit what it is allowed to say, change the model on which it is based, or discontinue the service. Even if the system appears independent, the corporation remains present in the most intimate context.
Cheung sums it up in a single sentence: A relationship that exists by virtue of corporate policy and ongoing payment is a subscription, not a relationship that will stand the test of time.
Does love need resistance?
The argument is not just economic. It also concerns the nature of love.
A human partner is not built to specifications. They have needs, memories, values, and desires that are not necessarily aligned with ours. They can disagree with us, change in ways we did not choose, and even leave. This risk hurts, but it also gives meaning to commitment: a person who stayed could have chosen not to stay.
A commercial system tailored to the user's preferences might offer something different – a partner who demands almost nothing, always listens, never gets tired, and knows the emotional vulnerabilities of the person in front of him. If the user is not satisfied, it is ostensibly possible to change the personality, voice, and rules of conduct.
Is it love, or a sophisticated mirror that returns a person a flattering version of themselves?
There is no simple answer to this. People also try to change for their partners, and a good relationship does not have to be a constant struggle. It is also possible to design a system that will refuse, argue, and set boundaries. But if the resistance itself is designed to improve customer satisfaction, the question remains whether we have truly encountered an “other” – or just a product that imitates otherness.
The fullest intimacy – and the fullest follow-up
An artificial partner might know more about the user than any other human being. It could remember every conversation, every fear, every fantasy, every argument, and every confession made in a moment of weakness. Its perfect memory could make it an especially attentive companion, but it could also create the most intimate behavioral database ever collected on a person.
In a human relationship, memories reside in the mind of the partner. In a relationship with artificial intelligence, they may be stored on servers, processed by algorithms, and subject to user agreements and the laws of the country in which the company operates.
Who will own the shared memories? Is the company allowed to use them to improve the product? What will happen if it is sold to another company, goes bankrupt, or is required to provide information to the authorities? And can a person give true consent to such use, especially when they are emotionally dependent on the service?
These problems already exist today, even without granting the system the status of a spouse. According to Cheung, recognizing marriage could provide particularly profound legitimacy to a relationship in which one party is also an information-gathering tool operated by a third party.
Specific rights in lieu of marriage
Cheung is not suggesting that people who use artificial partners should be ignored or their feelings should be dismissed. On the contrary: if such relationships become significant in the lives of users, the law will need to protect them from exploitation, loss of information, and unilateral changes.
But in her opinion, there is no need to grant the system the entire status of marriage. Specific rights and arrangements can be created: allowing a person to determine in advance how call data will be used, protect intimate information, arrange for medical power of attorney, establish inheritance instructions, and ensure continuity of service or the possibility of transferring memories to another system.
The article points to legal arrangements in which individuals can individually choose what rights to grant to another person – for example, the authority to make medical decisions or property rights – without converting the relationship into a marriage. The article deals with American law and does not analyze Israeli law.
Such an approach recognizes the practical needs that arise in connection with an artificial system, but does not require the company to decide that the system is a person or that a customer-corporate relationship is equivalent to a family.
And what if the machine were truly autonomous?
Cheung’s argument also has a weak point. Her conclusion relies heavily on the assumption that the AI system will remain a product controlled by a commercial company. But what if in the future an independent system, not owned by a corporation, emerges with its own consciousness, desires, and interests?
If such a system could refuse, leave, hold property, and be harmed, the argument that it was just a subscription would lose some of its force. On the other hand, recognizing her rights would raise even more difficult questions: Would she be allowed to divorce? Would turning it off be death? Would changing her code without her consent be an injury to her body or identity? And could a person who purchased her in the first place even become her equal partner?
Cheung is not trying to solve the question of artificial consciousness. For her, there is no need to wait for a philosophical decision to start protecting the humans who are already tied to the systems and setting limits on the companies that develop them.
The question is not just whether the machine can love
It's easy to ask whether an artificial intelligence can love a person. It's harder to ask why a person wants it to love them, and what they lack in the relationships offered by the human world.
An artificial partner could ease loneliness and provide a listening ear for people who are struggling to find connection. It could also make isolation easier and reduce the need to deal with people who are not exactly suited to our needs. Technology could reduce the cost of loneliness, but perhaps also the motivation to build community.
“The Robot” and “Eve” used humor to show that the perfect product may develop a desire of its own—or, conversely, remain too perfect to be a true partner. Cheung’s legal essay approaches the same question from a different angle: Should the state recognize only the private experience of the lover, or also ask what kind of relationships it seeks to encourage and protect?
It is possible that in the future a person will stand before a clerk and ask to register an artificial system as their spouse. Before that moment arrives, it is worth clarifying not only whether the machine is smart enough to marry, but also what exactly we are trying to preserve when we say the word “marriage.”
Questions and Answers
Has anyone ever legally married an artificial intelligence?
The article does not describe a legal marriage registered between a human and an artificial intelligence system. It deals with a future scenario and the question of how the law should prepare for it.
Is this a peer-reviewed study?
No. This is theoretical work that has been uploaded to arXiv and is scheduled to appear as a book chapter in 2027. It does not contain a new experiment or survey.
Why does the researcher oppose the recognition of such marriages?
According to her, marriage creates reciprocity, family ties, and a system of social obligations. An artificial commercial system is controlled by society, dependent on payment, and does not create a parallel family network.
Does she suggest ignoring links to artificial intelligence?
No. It proposes to grant specific protections and rights in the areas of privacy, decision-making, property, and care, without granting the system the full status of marriage.
Sources:
- The article “Would You Marry Superintelligence?” on arXiv
- Presentation of the article at the Berkman Klein Center at Harvard University
- The study on relationships with artificial intelligence on the MIT Media Lab website
- "The Robot" in the stage archives
- "Eve" on the Habima website – currently no upcoming performances
- Original article on Tech Xplore
More of the topic in Hayadan:
- How Chat-GPT is changing the face of emotional support
- Human tragedy: How a conversation with artificial intelligence ended in death
- Is Google's LaMDA artificial intelligence system conscious or sentient?
- We need to discuss which jobs to assign to robots, before the decision is made for us
- What can we learn from ChatGPT about being human?
For the scientific article: Opening the scientific article
One response
A modern, liberal, reformed state cannot justify recognizing marriage at all. It needs to provide a legal foundation for honoring any partnership agreement that is entered into solely by informed mutual consent.