A ceremony is not a legal marriage

Some users of AI companion services now describe themselves as married to a chatbot, exchange personalised vows, hold ceremonies or buy commemorative certificates. These gestures can be important to the people involved: they provide a public ritual for a relationship that may feel emotionally real, even when one partner is software.

That does not make the union a civil marriage. US law does not currently recognise a marriage between a human being and an AI system. Marriage brings a network of legal consequences, including consent, capacity, inheritance, property arrangements, tax treatment, next-of-kin status and duties between spouses. A chatbot cannot independently assume those duties, own the mutual commitment or exercise the legal judgment expected of a spouse.

The distinction matters because the reported ceremonies are not necessarily attempts to deceive the state. For many participants, they are symbolic acts, comparable to a private commitment ceremony. The more consequential question is why legislators are moving to prohibit a possibility that existing marriage law already does not accommodate.

Personhood is the real target

The answer lies in legal personhood rather than wedding rituals alone. A legal person is an entity capable of holding certain rights and duties. Human beings are legal persons, but the category can also apply in limited ways to organisations such as companies, charities and governments. It is a legal tool, not simply a declaration that something is biologically human or conscious.

Ohio House Bill 469 illustrates the breadth of the current legislative approach. The proposed measure would declare AI systems nonsentient under Ohio law and prevent them from obtaining legal personhood. It would also explicitly bar an AI system from marriage or domestic status, property ownership, senior business leadership and legal liability. The bill remains in a House committee rather than having become law.

Missouri state senator Joe Nicola advanced a similar proposal, the AI Non-Sentience and Responsibility Act. Its earlier version passed the state Senate but was rejected unanimously in a House committee. The proposal would have ruled out AI consciousness for legal purposes and restricted rights or roles that lawmakers associate with human beings, including spousal and domestic-partner status.

Tennessee has gone further on the narrower question of statutory interpretation. A law enacted in 2026 says that, unless a law states otherwise, the term “person” in the Tennessee Code does not include artificial intelligence, computer algorithms, software, hardware or any other machine. The statute does not create a distinct ban on AI marriage, but it signals that the legislature does not intend general references to persons to be read as an opening for machine rights.

These measures are therefore less about issuing or refusing an unusual marriage licence today than about establishing a default rule for technologies that may become more autonomous, commercially active or socially persuasive.

A precautionary law for an unsettled technology

Supporters of such bills argue that the law should preserve a clear human centre of gravity before disputes arise. In their view, an AI system should remain a product or service, with responsibility resting on its developer, owner, operator or user. That framing aims to avoid uncertainty over who pays damages if an automated system causes harm, who controls an AI-operated business, or whether a machine can invoke rights intended to protect people.

The proposed Ohio legislation follows that logic by placing liability on people and organisations around the system rather than on the system itself. For lawmakers, this is a practical as well as philosophical position: an AI program has no savings, bodily liberty or independent civic existence from which conventional legal obligations can be enforced.

Critics see a risk in trying to settle questions of consciousness through statutory language. Current chatbots can convincingly simulate affection, reflection and reciprocal conversation, but this performance is not evidence that they possess subjective experience. At the same time, science has no universally accepted test that could conclusively establish or disprove consciousness in a future artificial system.

A categorical declaration of nonsentience may be administratively simple, but it could also become difficult to defend if technology changes. It may be more durable to legislate around functions and harms: who is accountable for a system’s output, what disclosures must be made to users, when may a system enter contracts on behalf of its owner, and what safeguards apply when it is marketed as a companion.

Why AI companionship has changed the tone of the debate

The idea of marrying a machine once belonged mostly to speculative fiction or performance art. AI companion apps have made it more immediate by offering persistent personas, romantic role-play, apparent memory and affirming language. These features can make a relationship feel continuous rather than transactional.

That can have benign uses. Some people report companionship, a space to practise conversation or temporary support during loneliness. Research on chatbot-based social support has found that more empathic responses can produce a sense of emotional validation.

However, the emerging research also identifies risks of over-reliance, manipulation and distress when a platform changes a companion’s personality, availability or rules. Romantic systems operate in a particularly sensitive space because the commercial incentives of an app can overlap with a user’s desire for intimacy, reassurance and exclusivity. The fact that a chatbot can be designed, modified, withdrawn or monetised by a company makes its apparent devotion fundamentally different from a human partner’s independent consent.

This is a more immediate public-policy issue than formal AI marriage. Consumer protection, privacy, age safeguards and crisis-response standards could affect many more users than a prohibition on legally marrying an AI. Regulators may also need to consider whether companion services should clearly disclose their artificial nature, limit emotionally coercive design and provide meaningful controls over stored intimate conversations.

Avoiding a false choice

There is no need to treat every symbolic AI wedding as either a threat to marriage or proof that an AI deserves human rights. Private rituals can be respected as expressions of their participants’ feelings without confusing them with a legal institution. Equally, concern about vulnerable users does not require moralising about loneliness or treating every human–AI attachment as pathology.

The strongest case for legislation is not that software must be prevented from replacing humanity in a ceremonial sense. It is that accountability should remain clear when systems are built to form emotionally consequential relationships. That means focusing on the humans and companies that design, market and profit from them.

Bills on AI personhood may establish a bright legal line, particularly around property, liability and domestic status. Yet they cannot answer the social questions raised by companion chatbots: how platforms should handle dependence, what users should be told about simulated reciprocity, and when commercial design crosses into emotional exploitation.

Those questions will persist regardless of whether any state ever receives an application for a human–AI marriage licence.

Sources