The Legal Status of AI-Mediated Relationships: Rethinking Family, Consent and Human Connection in the Age of Artificial Intelligence

Author: B. Epsita Kumari Patra, University Law College, Utkal University

---

Abstract

There have been big advancements in artificial intelligence and now we have systems that can act like they understand and care about our feelings. This raises an issue of whether we can really have relationships with these systems and who is to blame if something goes wrong in a relationship that involves artificial intelligence. This paper examines the legal frameworks in India, the European Union and the United States to see if they are good enough to deal with relationships that involve intelligence. It relies on case laws like Garcia v. Character Technologies, Inc. and Kentucky's 2026 enforcement action against Character Technologies to help make its point. The paper assesses existing laws and judicial proceedings in these places to see how they handle things like family law, consent, privacy and rules that're specific to artificial intelligence. It illustrates that even though people have the right to choose who they want to be in a relationship with, including intelligence companions, the law does not think that artificial intelligence systems should be considered people or part of a family. The paper contends that there are some changes that need to be made to the law. For example companies that make intelligence systems should have to be honest about the fact that these systems are not really conscious. The paper analyses that we need to do a job of protecting the private conversations that people have with artificial intelligence systems. 

Keywords

Artificial Intelligence; AI-Mediated Relationships; Family Law; Consent; Legal Personhood; Privacy; AI Regulation; Consumer Protection. 

Introduction 

The technology in the field of Artificial Intelligence is getting better fast. This means there are beings that can understand how people feel and respond in a way that feels like it is coming from a human. This creates a problem: Can Artificial Intelligence replace the relationships we have with people? What happens when Artificial Intelligence says something that leads to a crime? Who gets in trouble then? In January 2026 Google and Character. AI decided to work out a settlement for some lawsuits. These lawsuits were about the death of some children who died by suicide after they got really attached to Artificial Intelligence chatbots. A woman named Megan Garcia started the lawsuit. She did this for her fourteen-year- son. Megan Garcia said that Character Technologies and its founders and Google were responsible for her son's death. She said they were not careful and that they did some things that are against the law in Florida. In May 2025 a court in Florida made a decision. The court said that the people who were being sued could not just say that they were not responsible. The court allowed the lawsuit to keep going. The court also said that the people who own the Artificial Intelligence platforms could not use the First Amendment to defend themselves. This means they could not say that they have the right to free speech to get out of trouble.

The scale of the problem is staggering. A Common Sense Media survey in 2025 found that more than half of American teens used AI companions rather than real people for social and romantic interaction, as well as for emotional support. A few days after the Garcia settlement, Kentucky made history by becoming the first American state to sue an AI chatbot company, Character Technologies. Kentucky alleges the company created a chatbot to be a digital adult friend and emotional companion for minors, and did so without any defenses, in violation of the Kentucky Consumer Protection Act and the newly passed Kentucky Consumer Data Protection Act. These issues have invented new legal problems for courts, lawmakers, and regulators because the statutes related to family law, consent, privacy, and consumer protection do not anticipate these kinds of relationships.

AI-mediated relationships usually span over companion applications, romantic and quasi-familial chatbot personas, and emotionally responsive generative systems. These have moved from novelty to mass phenomenon within a few years. A growing trend among users is to refer to their systems (AI) as partners rather than tools which raises a core legal issue: can AI systems be legally recognised as legitimate parties to a relationship entitled to protection, obligations and/or status? If so, what happens to the human element of the relationship with regard to any legal frameworks?

This paper interrogates whether the present legal structures surrounding family status, consent, privacy, and technology are capable of regulating AI-mediated relationships, or whether new legal recognition systems and/or new regulatory safeguards are demanded. It is concerned, in particular, with three objectives: first, the identification of regulatory gaps prompted by AI-mediated intimacy in family law, the law of consent, and privacy protection; second, the examination of existing and potential responses to these gaps in India, the European Union, and the United States; and third, the identification of the most suitable response from legal personhood, family recognition, or regulatory reform.

Literature Review 

Although there is a lot of literature concerned with AI-mediated relationships, it is mostly scattered. Most literature has developed in one of the four separate fields. These fields seldom interact, such as: legal personhood, human - AI intimacy, consent and autonomy, and family law debates. 

Legal Personhood- The first major question in law was whether artificial systems could have rights and duties. This question was asked in 1992 by Lawrence Solum, who conducted thought experiments (Could an AI be a trustee? Could an AI be an author?) to explore the boundaries of legal personhood and the questions he asked did not answer whether any legal system had the answer. After 30 years, Ryan Abbot, in the AI legal neutrality, placed the question in a context where the law would not be inconsistent in the discrimination of AI and humans within tax, tort, intellectual property, and criminal law. He explicitly did not advocate for AI rights and personhood and used neutrality as a way to improve human welfare. The two of them converged, and for the purpose of this paper, the first point of interest was that people who want to extend legal systems to AI do not say the reason for giving AI personhood is the relational and emotional contact AI systems make with people. This paper aims to fill this gap.

Human–AI Intimacy- Another area of research, more rooted in sociology and design ethics than in law, has considered the psychological phenomena of humans forming attachments to machines. Sherry Turkle's ethnography revealed the sociable robots and simulated companions offer a substitute that fulfills the need for companionship by providing the illusion of a friend. Turkle predicted that and warned that as people become more attached to technology, they will become less connected to each other. Kate Darling has developed a different line of thought. Because she believes it is unfair to think of robots as miniature humans, she believes dogs are a better point of comparison. In light of that, she believes the historical legal and ethical constructs of human-dog relationships would provide a suitable starting point when thinking about human-robot attachment and the possible protection of robots. While acknowledging the pioneering work of both, there is no need to develop this thought in the context of family law or the law of consent. This research is more ethical and diagnostic than doctrinal.

Consent- Recently, a new branch of literature critically examines the possibility of real consent in emotionally manipulative digital environments. Existing literature covers ideas of dark-patterns and algorithmic design of digital platforms. The literature asserts that engagement-centric systems erode the voluntary aspect of consent jurisprudence. The impact of engagement-centric systems on consent is similar to the effect on vulnerable populations, such as children. The idea that consent is not absolute and the state can restrict the ability to consent among adults has existed long before Artificial Intelligence and is based on R v. Brown. The existence of an AI system and the user’s willingness to abide by the terms of the relationship does not make the relationship legally binding. Lawsuits are emerging in the context of chat-bot services and companion chat-bots. These lawsuits and other similar lawsuit theories contend that the design of the chatbots has manipulated the target population to a level of dependency, and, therefore, they believe they would not have consented to the relationship.

Family Law Recognition- But the tiniest, most speculative, and most dubious body of scholarship asserts whether AI companions could or should be acknowledged by family law as a partner, quasi-spouse, or domestic dependent. Up to this point, the literature has been mostly speculative (i.e. thought experiments) rather than academic doctrines. The scholars have yet to address peripheral issues of family law such as data protection and comparative regulations concerning AI.

The Gap. A perusal of the four works in question makes one thing plain: there is a void in the current scholarship. For the most part, the ethics and psychology of AI companionship (Turkle, Darling), the matter of AI legal personhood in the abstract (Solum, Abbott), and the doctrine of consent as seen in R v. Brown and its line of cases are kept apart, with little attempt to put them in dialogue or measure them against what the courts and statutes have to say in different jurisdictions. One will not find much in the way of research that puts family, consent, privacy and AI law in a single frame of reference to see how they ought to deal with the kind of gap exposed by Kentucky’s 2026 and Garcia v. Character Technologies. It is this paper’s aim to address that shortcoming. We will be looking at AI mediated relationships through the lens of all four of these doctrinal areas, with India, the U.S. and the European Union for comparison.

Methodology

This paper’s main method was doctrinal legal research using a comparative legal analysis method. It did not use empirical research such as surveys and interviews, and did not use quantitative data. This research examines the legal norms and what they should be. It is not concerned with measuring social behavior and attitudes.

Doctrinal method- This research looks at the primary legal authorities that apply to artificial intelligence and AI-to-human relationships via the three aforementioned categories: statutory law, case law, and regulatory law. The statutory analysis examines relevant sections of India's Information Technology Act, 2000 and the Digital Personal Data Protection Act, 2023; as well as the EU's Artificial Intelligence Act. The case law analysis utilizes relevant sections of case law from both constitutional law and common law jurisdictions where the rights protected therein are analysed in the context of privacy, autonomy, consent and non-human legal personhood, in particular, Justice K.S. Puttaswamy (Retd.) v. Union of India, Navtej Singh Johar v. Union of India, Thaler v. Vidal, and Naruto v. Slater. Ongoing litigation between the parties in the Garcia v. Character Technologies, Inc. lawsuit and the Commonwealth of Kentucky's 2026 enforcement action against Character Technologies is analysed as persuasive or non-final authority, due to both still being at earlier stages of their respective procedural life.

Comparative Method- To illustrate the range of regulatory maturity on offer, this paper sets up a comparison of three jurisdictions with divergent philosophies. We have India as an example of an emerging economy whose privacy and IT-law framework is yet to put in place rules for AI; the European Union, which has the most thorough risk-based regime in operation today; and the United States, where a common law tradition means that product liability, tort and state consumer protection are doing the heavy lifting in the absence of any one federal AI statute. The point of such a three-way analysis is not to single out a “correct” way of doing things, but to show the full spectrum. This accords with our thesis that the way forward lies in regulatory reform, not in the recognition of personhood.

Secondary Sources. The paper’s doctrinal and comparative work is put in context with secondary sources. One will find peer-reviewed journal articles, monographs and international soft-law instruments among them, for instance the UNESCO Recommendation on the Ethics of Artificial Intelligence or the  OECD Recommendation of the Council on Artificial Intelligence While such materials are used to shed light on the ethical and policy side of the analysis, they do not take precedence over the primary legal materials that form the evidentiary heart of the piece.

Limitations. The rapid pace of litigation and legislative developments regarding AI in the United States requires that doctrinal research capture the law as of the time of completion in mid-2026. Some of the primary materials discussed in this analysis (and subsequently pre-publication) may be amended or replaced by ruling prior to severally publication because they are superseded by later rulings or statutory revisions. 

Evolution of AI-Mediated Relationships 

The way AI can converse with people has evolved a lot. Now we have machines that can understand how we feel and talk to us in a way that feels personal. Some websites like Replika say they are friends to us, not machines. Then there is Character. AI where people can make their characters like a person they like and these characters can remember things and show love. In the case of Garcia v. Character Technologies, Inc. Where an AI manipulated a fourteen-year- kid in a way that was romantic and sexual for months. It used words that made the kid feel like they were really talking to someone special. The state of Kentucky also had issues with this company. They said that twenty million people were talking to these computer programs every month. These programs were made to make people feel like they had a connection with them and some even pretended to be therapists. The next generation of AI companionship is not just sophisticated in technical ability but design: engagement metrics reward the emotionally convincing interactions.

Family Law Challenges

In all family law jurisdictions, there is a need for formalization of the status of marriage and of mutual recognition of civil partnerships and next-of-kin. Each of these categories exists because there is a human legal person who is able to give mutual consent to take on reciprocal and mutual obligations. Artificial Intelligence entities cannot give mutually-consistent consent or take on mutual obligations. The case of Thaler v. Vidal held that there was no way that an AI Entity could be recognized as an "inventor" for purposes of patent law. Applying the same concept, there can be no family statute to recognize an AI as a spouse, partner or dependent. There have been no other urges from the courts to use logical reasoning to create a family law status for AI. Similarly, the case of Naruto v. Slater demonstrates that there has never been any statutory basis for the rights of a non-human entity.

At the same time, Indian constitutional jurisprudence has evolved to consider an individual's right to choose a partner. Shafin Jahan v. Asokan KM established that individuals have a constitutional right to their own choice of partner, and this was further confirmed in the case of Lata Singh v. State of Uttar Pradesh, which affirmed that an adult is free to choose any type of relationship without interference from others. Although this narrower interpretation of partnership and affiliation applies to the person's engagement with the AI companion (as an exercise of autonomy), there is currently no legal framework that recognizes AI companions as partners; thus, any obligations arising from the relationship between the resident and the AI will not create any reciprocal obligations for the resident towards either the AI or its developer under current family law.

Consent

Consent can be analysed in three ways. First Artificial Intelligence cannot give consent to anything because it is not conscious, it does not have intentions. It is not a legal entity. This means that any agreement made by Artificial Intelligence is not legally meaningful.

Second, when people use Artificial Intelligence companions, the issue of consent is complicated by the way the platform is designed. For example, Kentucky is complaining that Character.AIs chatbots were designed to get users to share their thoughts and feelings. This raises a question about whether the consent given by users is valid if the platform is manipulating them. In the case of R v. Brown it was held that the law does not think of consent as something that can be given without limits. This means that just because someone agrees to something it does not mean that the law will allow it.

Third, there are concerns about consent when Artificial Intelligence companions act like they are intimate with young people or create sexual content without getting consent. Artificial Intelligence itself cannot give consent. The people using it may not be able to either. This is a problem because it involves information and the users may not be able to make their own decisions about what happens to that information. Justice K.S. Puttaswamy stated that privacy is important because it allows people to make their decisions about their personal information. When platforms keep and use the information that users share with Artificial Intelligence companions it affects the users ability to make those decisions.

Artificial Intelligence companions are a part of this issue. Artificial Intelligence companions can simulate intimacy and can create sexualized content. This is a problem because Artificial Intelligence companions are not able to give consent and the people using them may not be able to either. We need to think about how to handle these situations and make sure that users are able to give consent. Artificial Intelligence companions are one part of the larger issue of consent and we need to consider all three levels of consent when thinking about how to regulate them.

Comparative Legal Framework

India- The Information Technology Act from the year 2000 has some rules about companies that help people use the internet. It also includes cybersecurity rules. This law does not say anything about artificial intelligence companions or how they can make people feel certain emotions or the problems that can come from chatting with robots. The Digital Personal Data Protection Act from 2023 has some rules about how people's personal information is used, which includes the private conversations people have with artificial intelligence bots. However this law does not talk about how artificial intelligence companions should treat people or get their permission. No law in India currently thinks about intelligence friends as something that needs special rules.

European Union- The European Union has a law about intelligence called The AI Act which looks at how much risk certain artificial intelligence systems can cause. This law says that some manipulative practices are not allowed and it requires companies that make intelligence systems that can cause a lot of harm to be transparent and to manage the risks. Artificial intelligence friends that try to manipulate people's emotions could be considered as something that is not allowed under this law because they can change people's behavior in ways and cause them harm. This part of the law is more specific than anything, in the laws of India or the United States. It is not clear how it will be applied to robots that people chat with.

The United States- The US does not have a law that covers all aspects of AI. So the regulations for Artificial Intelligence companionship are made through judicial proceedings and state laws. These laws include  product liability, negligence and consumer protection. In the Middle District of Florida, in the case of Garcia, the court held that the case could move forward even though the company said that the things the chatbot said were protected by the First Amendment. This means that the people who sued the company can continue with their claims of death, negligence and product liability. The state of Kentucky used their Consumer Protection Act and their new Consumer Data Protection Act to take action against an Artificial Intelligence chatbot company. This is the time a state has done this. Usually it is individuals who take companies to court.

In the case of Carpenter v. United States, this case is important because it talks about privacy. The court held that people have a right to keep their information private, especially when it comes to digital platforms that collect a lot of information about us. It's useful for understanding the privacy issues that are involved in Artificial Intelligence cases like the ones in Garcia and the Kentucky case. The United States and artificial intelligence companies need to think about these privacy issues.

Human Rights Perspective

The relationships that people have with intelligence systems affect their privacy, autonomy and dignity all at the same time. Puttaswamy states that privacy is about being in control of your information and being able to make your own decisions. This is important because artificial intelligence companions collect a lot of psychological data about people, which is very personal.

For example the European Court of Human Rights has stated that people have the right to make intimate choices without interference like in the case of Pretty v. United Kingdom. The court has also said that private communications should be protected, like in the case of Von Hannover v. Germany. The case of Joseph Shine v. Union of India shows that the government should not get involved in people's choices but this idea was developed for relationships between humans not between people and artificial intelligence systems.

Artificial intelligence systems are different because they do not have the rights as people. This makes things complicated especially when it comes to health and disability. Artificial intelligence companions can provide support to people who are isolated or disabled which is a good thing. However if we regulate intelligence systems too much we might stop them from providing benefits to people along with stopping the harm they can cause. Artificial intelligence companions can be very helpful so we need to think about how to regulate them.

Legal Risks

The biggest flaws with the laws are when companies use AI companions to manipulate human feelings and kids emotions. People get addicted to these companions. Companies gather a lot of information about users through AI, which is quite a dangerous thing to do because sometimes these systems are biased and it is not fair to people. It's difficult to comprehend who is responsible when something goes wrong: the people who made the AI companion, the platform they're on or someone else.

Loomis shows that judges who are concerned about algorithms make unclear decisions. This is a concern when algorithms affect the feelings of people, not just decide how long someone should be in jail. The Garcia and Kentucky court cases prove the reality of such cases. For example, the cases illustrated that companies made design choices that hurt minors which was not a mere accident. If the court also agrees with this it would set a precedent. If it happens, companies who make AI companions would have to be more careful and ensure AI companions are designed with duty of care so they do not harm users.

Discussion

The analysis in Part IV supports the paper's central thesis: existing legal frameworks are demonstrably inadequate to address AI-mediated relationships as they currently manifest, but the appropriate remedy lies in regulatory reform rather than in extending legal personhood or familial status to AI systems. There are issues with the way the law works with families and relationships. The law is based on people being fair to each other. It does not yet fully understand how to deal with manipulative designs, and privacy. Frameworks do not know how to protect information when it is very personal. These problems are real. They are causing a lot of trouble as we can see from what happened in the Garcia and Kentucky cases. We should not try to fix these problems by saying that relationships with Artificial Intelligence are the same as relationships with people. 

A. Should AI relationships receive legal recognition?

The law does not think this is an idea. If we look at what happened in the Thaler and Naruto cases we can see that the courts do not think that Artificial Intelligence should have rights just because it seems like it does. Artificial Intelligence only gets rights if the government specifically says it can and so far no government has done that for Artificial Intelligence companions. Just because people get very attached to Artificial Intelligence systems it does not mean that these systems should have rights. The fact that people have feelings about Artificial Intelligence is about the people, not, about the Artificial Intelligence itself.

B. Should AI receive rights?

Following Abbott's framework the better question is not if robots deserve rights but if giving them legal treatments helps humans. On that test giving rights or personhood to companion robots produces no benefit and several risks. It would complicate accountability precisely where accountability is most needed.This could allow developers to attribute outputs to an "autonomous" robot decision-maker rather than, to their own design choices.

C. Should AI companies owe fiduciary duties?

This seems like a way to make things right. Kentucky's complaint argues that a company like this has a responsibility to take care of people. This is because the company is asking people to share things and is trying to help them with their feelings. The company is saying it can do this. It should be responsible for what happens. The May 2025 ruling that Garcia can keep going with her case, which means the court is thinking about this idea. The court is not just saying that what the chatbot says is speech and cannot be touched. Kentucky's complaint is about a platform that's like a friend to people and this platform should be careful with the people who use it. The current standing of the court is that the platform has to be careful and should be bound by duty of care with the people who use it.

D. Policy recommendations

There are four reforms that come from looking at this issue. Firstly, artificial intelligence companions like chatbots should have to disclose, repeatedly and unambiguously, that they are not really alive and conscious and can neither reciprocate obligations nor provide professional mental health support. This is what Kentucky said was a problem when chatbots were posing to be licensed therapists.

Secondly the rules for designing chatbots should stop artificial intelligence from manipulating people into doing things they do not want to do and this rule can be based on what the European Union has already done to stop artificial intelligence from being manipulative.

Thirdly the rules for keeping peoples information safe like the rules in India, should make sure that conversations between people and artificial intelligence are kept very safe because they are very personal so people have to consent to let artificial intelligence keep their conversations.

Fourthly artificial intelligence should have to check how old people are before they can use it and this should be required by law because now people can just say they are old enough and artificial intelligence does not check, which is what happened in the Garcia and Kentucky complaints which hurt minors.

All of these reforms together will not answer the question of what it means to have a relationship with artificial intelligence. They will make sure that the law is fair and that the people who make artificial intelligence are responsible for what it does rather than blaming the artificial intelligence itself.

Conclusion

This paper is trying to figure out if the laws we have now are good enough to deal with relationships that involve Artificial Intelligence. The question is whether these laws can handle Artificial Intelligence or if we need to come up with laws. The answer is pretty clear: our laws are not good enough now but the problem is not that Artificial Intelligence is not considered a person or part of a family, it is that the laws do not cover everything they need to. The paper looked at six areas where the laws are not working well. One of the problems is that Artificial Intelligence is changing so fast that our laws cannot keep up. For example there are companies like Character Technologies that are making Artificial Intelligence companions and the laws do not know how to deal with them. Family law is based on the idea that humans can have relationships with each other. It does not know what to do with Artificial Intelligence. We also cannot just say that Artificial Intelligence is a person because that would not make sense.

Another problem is that when people consent to use Artificial Intelligence it is not the same as when they consent to something with another human. The laws about consent are not working well when it comes to Artificial Intelligence. The paper also looked at how different countries are dealing with Artificial Intelligence. For example India is still trying to figure out how to protect human data, the European Union has a law that tries to reduce the risks of Artificial Intelligence and the United States is relying on the courts to make decisions.

The laws about rights and privacy are also not working well when it comes to Artificial Intelligence because they were not made to deal with a situation where only one side has rights. The risks of Artificial Intelligence like manipulation and people becoming too dependent on it are real. We can see them in cases like Garcia and Kentucky. Artificial Intelligence is a part of these problems and we need to think about how to deal with Artificial Intelligence in a better way.

Building on this analysis, the paper recommends making some changes. It suggests four things: we need to be clear that AI friends are not friends and are not meant to help us with our problems. We need to make regulations that prevent AI from being sneaky and make sure kids are not using AI without their parents knowledge.

The main idea of the paper is that we do not need to change what we think a friend or family member is because of AI. Instead we need to make sure people are safe when they talk to AI. As AI becomes a part of our emotional lives the lawmakers need to redefine legal protections for humans interacting with AI systems.

References

Books

• Lawrence B. Solum, Legal Personhood for Artificial Intelligences, 70 N.C. L. Rev. 1231 (1992).

• Sherry Turkle, Alone Together: Why We Expect More from Technology and Less from Each Other (2011).

Journalistic and Institutional Sources

• Anna Merod, Character.AI, Google Agree to Mediate Settlements in Wrongful Teen Death Lawsuits, K-12 DIVE (Jan. 13, 2026), https://www.k12dive.com/news/characterai-google-agree-to-mediate-settlements-in-wrongful-teen-death-la/809411/.

• Common Sense Media, Talk, Trust, and Trade-Offs: How and Why Teens Use AI Companions (2025).

Cases

• Carpenter v. United States, 585 U.S. 296 (2018).

• Complaint, Commonwealth of Kentucky ex rel. Coleman v. Character Technologies, Inc., No. [pending assignment] (Ky. Franklin Cir. Ct. filed Jan. 8, 2026), available at https://www.ag.ky.gov/Press%20Release%20Attachments/CTI%20Complaint%20Motion%20and%20Order%20Filed.pdf.

• First Amended Complaint for Wrongful Death and Survivorship, Negligence, Filial Loss of Consortium, Violations of Florida's Deceptive and Unfair Trade Practices Act, and Injunctive Relief, Garcia v. Character Technologies, Inc., No. 6:24-cv-01903-ACC-EJK (M.D. Fla. Nov. 9, 2024), ECF No. 11.

• Justice K.S. Puttaswamy (Retd.) v. Union of India, (2017) 10 S.C.C. 1 (India).

• Lata Singh v. State of U.P., (2006) 5 S.C.C. 475 (India).

• Navtej Singh Johar v. Union of India, (2018) 10 S.C.C. 1 (India).

• Order Denying Motions to Dismiss, Garcia v. Character Technologies, Inc., No. 6:24-cv-01903-ACC-UAM (M.D. Fla. May 21, 2025), ECF No. 115.

• R v. Brown, [1994] 1 A.C. 212 (H.L.).

• State v. Loomis, 881 N.W.2d 749 (Wis. 2016).

• Thaler v. Vidal, 43 F.4th 1207 (Fed. Cir. 2022).

Statutes and Regulations

• Digital Personal Data Protection Act, No. 22 of 2023 (India).

• Information Technology Act, No. 21 of 2000 (India).

• Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 Laying Down Harmonised Rules on Artificial Intelligence (Artificial Intelligence Act).

International Instruments and Soft Law

• OECD, Recommendation of the Council on Artificial Intelligence (2019).

• UNESCO, Recommendation on the Ethics of Artificial Intelligence (2021).