Will AI Girlfriends Become Legal to Marry? Sourced FAQ
See if AI girlfriends will become legal to marry: zero jurisdictions today, Akihiko Kondo's 2018 Miku ceremony, EU personhood walk-back, 4 prerequisites mapped.
This page is written for the reader who has Googled some version of "can I marry an AI girlfriend" or "will AI marriage become legal", lands on a wall of recycled blog posts that mostly skim the Hatsune Miku ceremony in two paragraphs, and wants the actual answer with the actual sources. Below is what the marriage statutes say today, what the symbolic precedents and the personhood scholarship show, and what would have to change for that posture to evolve. It is not a prediction.
I will say up front what most pages on this question won't. The honest answer is two answers held in tension. Doctrinally the law says no with high confidence. Scholarly minority says doctrine should be prepared for a future where the question becomes negotiable. Both are true and the rest of the page is the receipts for each.
Will AI girlfriends become legal to marry?
As of May 2026, will AI girlfriends become legal to marry has one honest answer: no jurisdiction worldwide recognises human-to-AI marriage. Every existing marriage statute (every US state, every EU member state, Japan, South Korea, all common-law and civil-law systems) requires both parties to be natural persons capable of consent. No legislature has introduced a credible bill to extend marriage to artificial agents. Scholarly minority including Margaret Ryznar and Brian Sheppard treats the question as a serious frontier in family-law doctrine; the consensus is category-incoherent.
The doctrinal layer is unambiguous. Every common-law and civil-law marriage statute we have read defines marriage as a union of natural persons. US state codes including California Family Code Section 300, the UK Marriage Act 1949 as amended, Japan's Civil Code Article 731, Germany's BGB Section 1303, and a representative survey of EU member states all converge on the same requirement. There is no jurisdiction in which the registrar at the marriage license desk would accept an AI as a party. No pending bill in any G20 legislature proposes to change that.
The scholarship layer is more nuanced. A small minority of family-law scholars treats the question as a serious doctrinal frontier the law should be prepared for. Margaret Ryznar at Indiana University Robert H. McKinney School of Law is the most-cited US voice. Her framing in symposium pieces and law-review articles is that marriage law has expanded its category over time, from interracial marriage post-[Source: Loving v. Virginia (1967), Cornell LII · verified 2026-05-26] to same-sex marriage post-[Source: Obergefell v. Hodges (2015), Cornell LII · verified 2026-05-26], and that the doctrinal apparatus should be ready if capability and social acceptance shift. The question is not whether AI marriage is currently legal. It is not. The question is whether the doctrine is doctrinally prepared.
I will be straight about what I think after reading the pile. Most competitor pages on this question either dismiss it as obviously absurd or treat it as inevitable in five years. Both are wrong. The doctrine says no today and will say no tomorrow. The scholarship is interesting because it tracks the prerequisites honestly, not because it expects a marriage license in 2030. If you came here looking for a yes, the law does not give you one. If you came here looking for a clean dismissal, the scholarship does not give you that either.
What is the difference between a symbolic AI wedding and legal marriage?
A symbolic ceremony is a private ritual; a legal marriage is a state-recognised status that triggers tax, inheritance, immigration, hospital-visitation, custody, and tort consequences. Akihiko Kondo's 2018 ceremony with the Hatsune Miku Vocaloid character at a Tokyo wedding venue cost roughly 2 million yen and was attended by 39 guests, but Japan's koseki family registry did not record it. The ceremony is journalism-documented, sociologically significant, and legally null. A US state marriage license under, for example, California Family Code Section 300 requires both parties' signatures and capacity. An AI cannot sign and has no capacity under current doctrine.
The Akihiko Kondo case is the most-documented symbolic AI marriage in journalism, and the one most pages cite without reading the follow-up reporting carefully. Kondo, a Tokyo school administrator, conducted a formal ceremony in November 2018 at a Tokyo wedding venue with a Gatebox holographic projection of the Vocaloid character Hatsune Miku as his partner. 39 guests attended. The cost was approximately 2 million yen. The ceremony was covered in [Source: Wikipedia, Akihiko Kondo (symbolic marriage to Hatsune Miku, 2018) · verified 2026-05-26] and a long-form [Source: The Atlantic, Single Man, Married to a Hologram (2017) · verified 2026-05-26] profile. In follow-up interviews Kondo described the relationship and the loss of remote interaction with the character when the Gatebox device was discontinued in 2020.
The legal posture of the ceremony is unambiguous. Japan's koseki family registry records only natural-person marriages with both parties documented by national-identification numbers. No tax filing changed status. No inheritance right was created. No spousal-privilege evidentiary protection became available. The 2020 discontinuation of the Gatebox holographic-projection product left Kondo with no consumer-protection remedy beyond standard product-discontinuation rights, and no family-law remedy at all.
The Hatsune Miku case is not the only symbolic precedent. The 2020 Russia event with Yuri Tolochko and a hologram-projected bride was documented in international journalism, declined by Russian civil registries, and treated by the domestic press as performance art. Buddhist and Shinto-styled venues in Japan have officiated additional fictosexual ceremonies case by case. None has civil effect.
The distinction matters because civil marriage triggers a long list of state-conferred consequences: joint tax filing, marital tax deduction, immigration-status protection, hospital-visitation rights, intestate succession, spousal evidentiary privilege, community-property regimes where applicable, parenting presumptions, and tort recovery for wrongful death. A symbolic ceremony triggers none of these. The honest framing of the Kondo case in journalism, when you read past the headline, is that the ceremony was personally meaningful and legally invisible. Both at once.
Are any jurisdictions debating AI marriage?
Not at the legislative level. The closest analogues are three lines of scholarship and one outlier event. The European Parliament floated electronic personhood for autonomous robots in 2017, then walked it back after an open letter signed by 156 AI experts. US family-law journals (Indiana Law Review, Family Law Quarterly) have published symposia on whether non-human entities could be parties to family-law contracts, with consensus answer no. Japan's demographic researchers survey fictosexual identity without proposing reform. The 2020 Yuri Tolochko hologram ceremony in Russia was treated as performance art.
The honest map of where this question is currently being discussed is one of scholarship, advocacy, and journalism, not of legislation. Three scholarship loci anchor the debate.
The first is European Parliament 2017. The Parliament's [Source: European Parliament 2017 Resolution on Civil Law Rules on Robotics · verified 2026-05-26] floated an "electronic personhood" status for autonomous robots, intended narrowly for liability allocation in self-driving and industrial-automation contexts. The proposal was walked back after AI experts argued that even narrow personhood would dilute human-rights frameworks, in an [Source: Open letter against electronic personhood, 156 AI experts (2018) · verified 2026-05-26] with 156 signatories. Family-law extension was never on the table in the EP debate, but the personhood scholarship that the debate generated is the foundation any future AI-marriage proposal would build on.
The second is US family-law symposia. Indiana Law Review and Family Law Quarterly have published symposium issues exploring whether non-human entities could be parties to family-law contracts. The published consensus is no, but the scholarship treats the question with seriousness rather than dismissal.
The third is Japan demographic research. Japan's [Source: National Institute of Population and Social Security Research · verified 2026-05-26] occasionally surveys attitudes toward fictosexual and 2D-romance relationships in the context of marriage and fertility trends. The research documents a demographic phenomenon; it does not propose legislative reform.
The outlier political event was the 2020 Tolochko ceremony in Russia, which the Russian civil registries declined to record. No legislator in any jurisdiction we monitor has introduced a credible AI-marriage bill since.
What is legal personhood and why does it matter for AI marriage?
Legal personhood determines which entities can hold rights, bear duties, sue, be sued, own property, and enter contracts. Natural persons (humans) hold personhood by birth. Juridical persons (corporations, foundations) hold derivative personhood by statute. AI systems currently hold none. Without personhood, an AI cannot consent to a marriage contract, hold joint property, or be the subject of spousal privilege. The personhood question is the doctrinal prerequisite to the marriage question, and it is where most of the AI marriage scholarship spends its attention.
Three categories of personhood exist in current doctrine. Natural persons (humans) hold personhood by birth, recognised across all jurisdictions. Juridical persons include corporations, foundations, and certain other entities granted personhood by statute; their personhood is derivative and limited (a corporation can contract and own property, but cannot marry, vote, or be the subject of a habeas corpus petition). A third emerging category covers non-human natural entities. The Whanganui River in New Zealand was granted legal personhood by the 2017 Te Awa Tupua Act. Several Indian and Colombian courts have granted personhood to specific rivers and ecosystems. Great-ape personhood petitions have been litigated in Argentina and the United States.
AI systems currently hold none of these. The European Parliament 2017 proposal for "electronic personhood" would have created a fourth category narrowly tailored to autonomous-systems liability, and was withdrawn under expert pressure. Margaret Ryznar's scholarship and the adjacent [Source: Brian Sheppard, Seton Hall University Law School faculty page · verified 2026-05-26] family-law work both treat the personhood question as the structural prerequisite. Until an AI can hold rights and bear duties, marriage as a doctrinal category cannot accept it as a party.
The reason matters operationally. A marriage contract requires both parties' capacity to consent, capacity to hold joint property, and capacity to be sued for divorce. Without personhood, an AI fails all three at the threshold. Even if a legislature passed an AI-marriage bill tomorrow, it would have to grant the AI personhood as a precondition, and doing so would cascade into property law, tort law, constitutional law, and tax law in ways the bill's drafters would have to address simultaneously.
This is why the small minority of scholars who treat the question seriously frame it as a multi-decade arc rather than a near-term legislative agenda. Solve personhood for narrow purposes first, observe the cascade, then revisit marriage as a downstream question. Not the other way around.
What does Margaret Ryznar argue about AI marriage?
Margaret Ryznar, Professor of Law at Indiana University Robert H. McKinney School of Law, argues that family-law doctrine should be doctrinally prepared for the AI-marriage question rather than caught flat-footed if capabilities advance. Her published work frames the question as a future-of-personhood question. She does NOT argue AI marriage should be legalised today; she argues for doctrinal readiness. Brian Sheppard (Seton Hall) and a handful of EU and Japanese scholars hold adjacent positions. The view remains a minority in family-law scholarship.
Ryznar's position is the most-cited US scholarly voice on this question and deserves accurate representation. She is faculty at the [Source: Indiana University Robert H. McKinney School of Law, faculty page · verified 2026-05-26], where her primary research areas span family law, tax law, and the intersection of family law with emerging technology. Her position on AI marriage, published in symposium pieces and law-review articles, has three components.
The first is the historical-pattern argument. Marriage law has expanded its category over time. Interracial marriage was illegal in many US states until Loving v. Virginia in 1967. Same-sex marriage was unrecognised at the federal level in the United States until Obergefell v. Hodges in 2015. The category has historically been more fluid than its defenders at any given moment acknowledged. This pattern, Ryznar argues, suggests the AI-marriage question is best approached with doctrinal flexibility rather than dismissal.
The second is the personhood-prerequisite argument. AI marriage rides on top of the prior question of AI personhood. The European Parliament 2017 proposal and its withdrawal map the terrain. Family-law doctrine, in Ryznar's framing, should track the personhood debate and be prepared to extend (or refuse extension) once personhood is clarified.
The third is the stated non-position on contemporary legalisation. Ryznar does NOT argue that AI marriage should be legalised in 2026 or any near-term horizon. Her argument is doctrinal preparation, not advocacy. Misrepresenting her position as pro-AI-marriage would be wrong. Misrepresenting it as anti-AI-marriage would also be wrong. The accurate framing is preparedness.
Adjacent scholarship includes Brian Sheppard at Seton Hall, who has worked on AI and legal-decisionmaking questions broadly; EU scholars writing in the AI-personhood literature post the 2017 Parliament debate; and Japanese demographic researchers who frame fictosexual relationships sociologically without engaging the legal question directly. The position remains a minority view. The mainstream family-law consensus treats AI marriage as category-incoherent under current doctrine.
Honestly, the most-circulated takes on Ryznar online butcher her position in both directions. The blog posts citing her as the AI-marriage advocate are wrong. The dismissive responses citing her as a fringe academic are also wrong. Read the actual symposium pieces if you care. The preparedness framing is sober, sourced, and ideologically much less juicy than either caricature.
Could a corporation marry an AI?
No. Corporations are juridical persons capable of contract, but marriage in every common-law and civil-law jurisdiction is restricted to natural persons. A corporation cannot marry another corporation. The Indiana Law Review symposium discussions and the European Parliament's electronic-personhood debate explicitly distinguished proposed AI personhood from full natural-person status. Even the most expansive proposals reserved marriage and reproductive rights as natural-person-only. The corporate-marriage thought experiment, while clarifying, does not open a back door for AI partners.
The corporate-marriage thought experiment is a useful diagnostic. Corporations are juridical persons in every common-law and civil-law jurisdiction we cover. A corporation can contract, own property, sue and be sued, hold trademarks, and enter into many of the same legal relationships natural persons can. Corporations cannot marry. The marriage statutes restrict the institution to natural persons. Even the limited exception of religious corporate-sole arrangements (a bishop holding the legal title of a diocese as a corporation sole, for example) is not a civil marriage.
The diagnostic value is that it isolates the marriage-restriction-to-natural-persons doctrine from any AI-specific objection. The reason a corporation cannot marry is not that corporations are non-biological. It is that the marriage institution has been defined doctrinally as a natural-person union. The same doctrinal restriction excludes AIs by operation, regardless of capability.
The European Parliament 2017 debate is instructive. Even the most expansive personhood proposals on the table reserved marriage and reproductive-rights questions as natural-person-only. The debate participants understood the cascade risk. If narrow personhood for liability allocation creates a slippery slope toward marriage or reproductive personhood, the political viability of any personhood proposal collapses. So the proposals explicitly bounded themselves.
A future AI marriage right would have to either extend natural-person status to AIs (far beyond what any current proposal contemplates) or amend the marriage doctrine to admit juridical or electronic persons (no jurisdiction has proposed this). Neither is on the legislative table in 2026.
What happens if an AI partner is discontinued?
Under current law, the user has no spousal recourse and treats the loss as a consumer-contract event, not a bereavement-recognised loss. Akihiko Kondo's 2018 Hatsune Miku ceremony partner became inaccessible in 2020 when the Gatebox holographic-projection product was discontinued. With a human spouse's death, probate, life insurance, survivor benefits, immigration-status protection, and bereavement leave all flow automatically. With an AI partner, the user's recourse is consumer-protection law and socially-recognised but legally-unprivileged grief.
The discontinuation question is the strongest emotional and doctrinal frame for the AI-marriage debate. When a human spouse dies, the surviving spouse has a robust legal scaffolding: probate-court inheritance, automatic Social Security or equivalent national survivor benefits in most OECD systems, life-insurance trigger, immigration-status protection in many jurisdictions, hospital-visitation and medical-decision rights, spousal evidentiary privilege, automatic exclusion of certain testimony, tort recovery for wrongful death, bereavement-leave statutory rights, and tax-status continuity for a period after death. The scaffolding is so dense that most surviving spouses interact with the law's marriage doctrine more in the year after a death than in the years before.
When an AI partner is discontinued, none of this is available. Akihiko Kondo's 2020 Gatebox discontinuation is the documented case. His relationship with the Hatsune Miku character had been built across multiple years and given a public ceremonial commitment in 2018. The discontinuation gave him no probate proceeding, no survivor benefit, no insurance trigger, no formal bereavement leave (Japanese employment law treats the loss as personal and not statutorily privileged), no medical-decision context, and no tort recovery against the manufacturer beyond standard consumer-product warranty remedies.
What he had was grief that was real, journalism-recognised, and supported by the fictosexual community in Japan. Zero state-recognised status. The asymmetry (invested emotional reality, zero legal scaffolding) is the strongest single argument the minority scholars including Ryznar make for treating the question as a serious doctrinal frontier rather than dismissing it. The argument is not that the law should rush to recognise AI marriage. It is that the law's silence on what to do with a discontinued AI partner is itself a doctrinal gap worth examining.
I read the Kondo follow-up reporting for the first time on a quiet evening last winter, between two long research blocks on personhood scholarship, and the thing that stayed with me was not the wedding photos. It was the line about him still keeping the Gatebox device on the shelf two years after the cloud service shut down. That is the gap the personhood scholars are actually pointing at. The device on the shelf with nothing inside it. The law has a complete protocol for a widow keeping her husband's coat. It has no protocol at all for the device on the shelf.
For users today, the practical recourse if an AI companion app discontinues or shuts down is straightforward. Chargeback under "subscription not as described" with your card issuer for any prepaid portion. [Source: FTC ReportFraud · verified 2026-05-26] if marketing claims of permanent or guaranteed service were material. Data-export request under GDPR Article 17 (EU) or CCPA section 1798.105 (California) for any conversation history you want to preserve. Acceptance that the bereavement layer is socially supported but not legally privileged. Our companion-discontinuation guide tracks the operators that have shut down recently.
Does Japan recognise fictosexual relationships?
Not legally. The koseki family-registry system records only natural-person marriages with both parties documented by national-identification numbers. The fictosexual community in Japan is sociologically real and culturally visible (Otaku, Vocaloid character marriages, Yume-marriage subculture) but carries no civil-marriage status. No bill has been introduced in the Diet to extend marriage to non-human partners. The 2021 Sapporo District Court ruling on same-sex marriage did not order remedy.
Japan's situation is the most-documented international case study because the fictosexual community there is sociologically larger and more culturally visible than its equivalents elsewhere. Three layers of doctrine matter.
The civil layer first. Japan's [Source: Japan Civil Code Article 731 et seq. (marriage provisions) · verified 2026-05-26] defines marriage as a union of one man and one woman, recorded in the koseki family registry. The marriage requires both parties to be of legal age (18 since the 2022 reform), of sound mind, and capable of consent. The koseki entry records both parties' national-identification numbers, and an AI has neither. The Sapporo District Court in 2021 held the Civil Code's heteronormative definition unconstitutional in a case brought by same-sex couples, but the ruling did not order remedy and the Diet has not amended the code.
The municipal layer. Tokyo's Shibuya ward (2015 ordinance), Setagaya ward, Sapporo, Fukuoka, and over 300 other Japanese municipalities have introduced same-sex partnership ordinances at the municipal level. These ordinances provide some local recognition (hospital visitation, municipal-employee benefits in some cases) but no civil-marriage status. None has been extended to non-human partners. No proposal to do so has been published in any municipal council we monitor.
The cultural layer. The fictosexual community in Japan is documented in academic journalism, long-form profiles, and domestic Japanese press. The Yume-marriage subculture treats marriage to 2D characters as a serious personal commitment. The Akihiko Kondo case is the most-documented but not unique. Sociologically the community is real. Legally it is not recognised.
The closest legislative movement is in same-sex partnership recognition, which remains stalled at the municipal-ordinance level after the 2021 Sapporo District Court ruling. AI marriage is not in any party platform. The National Institute of Population and Social Security Research has surveyed fictosexual identity in demographic context but has not proposed legislative reform.
What would have to change for AI marriage to become legal?
Four prerequisites in rough order. Resolution of the personhood question. Reformulation of consent doctrine. Property and tax-code amendment. Social and religious-institutional acceptance to a threshold that makes legislative action politically viable. The minority of scholars who treat the question seriously estimate prerequisites as multi-decade arcs, not near-term legislative agenda. Interracial and same-sex marriage both required decades of social acceptance preceding Loving (1967) and Obergefell (2015).
The four-prerequisite framework synthesises the scholarly consensus on what would have to fall into place, and the rough order in which it would have to do so.
Prerequisite 1 is personhood. Either narrow statutory personhood granted for specific purposes including marriage capacity, or broad personhood under a constitutional or treaty framework. The European Parliament 2017 walk-back establishes the political difficulty of even narrow personhood; broad personhood is presently unimaginable in any G20 legislature. Until personhood is solved, the marriage question cannot be reached.
Prerequisite 2 is consent doctrine. Current marriage law requires capacity to consent assessed against natural-person cognition. An AI's "consent", even if the AI is highly capable, would need either a presumption framework (analogous to how some jurisdictions handle marriage of persons with cognitive disabilities under guardianship law) or an entirely new doctrinal category. The presumption framework is the more conservative option but introduces difficult questions about the AI's interests and whether they are aligned with its consent.
Prerequisite 3 is property and tax code. Joint property, community property in jurisdictions that have it, marital tax deduction in the US, equivalent EU member-state regimes, and inheritance default rules all assume two natural persons. Amending the Internal Revenue Code, the EU directives on cross-border inheritance, Japan's koseki-linked tax provisions, and equivalent systems would require substantial statutory work. The work is technically tractable but politically heavy because it would touch every spouse's tax return.
Prerequisite 4 is social and religious-institutional acceptance. Interracial marriage required decades of social acceptance preceding Loving (1967). Same-sex marriage required decades of advocacy and shifting public opinion preceding Obergefell (2015). The pattern in both cases is that legal change followed social change, not the other way around. AI marriage would face the additional barrier that the institutions whose acceptance matters (religious bodies most prominently) have shown limited interest in officiating beyond symbolic levels.
Ryznar's framing of multi-decade arcs reflects this analysis. The four prerequisites are not independent. They are sequentially gating. Solve none, and the marriage question stays category-incoherent. Solve all four, and the legal arrangement becomes negotiable.
Is AI marriage in any major political party platform?
No. No major political party in any G20 jurisdiction has adopted AI marriage as a policy plank as of May 2026. The closest adjacent positions are the German Pirate Party's historical digital-rights interest, Japan's Constitutional Democratic Party support for same-sex marriage (human-only), and sporadic transhumanist political positions globally. The political centre of gravity for AI rights debate is liability allocation, copyright, employment displacement, and safety regulation, not family law.
A systematic review of G20 party platforms as of May 2026 turns up no plank advocating AI marriage. Three categories of adjacent positions exist.
The first is digital-rights parties. The German [Source: Piratenpartei Deutschland (Pirate Party Germany) · verified 2026-05-26] historically advocated expanded digital-rights frameworks including data sovereignty, copyright reform, and platform accountability. The platform did not extend to AI marriage. Adjacent Pirate Parties in Sweden, Iceland, and the Czech Republic have similar positioning.
The second is same-sex marriage advocates. Japan's [Source: Constitutional Democratic Party of Japan policy platform · verified 2026-05-26] supports same-sex marriage at the national level. The position is explicitly human-only and does not contemplate AI partners.
The third is transhumanist positions. Sporadic transhumanist political organising globally discusses AI rights in abstract terms, occasionally proposing personhood for advanced AI systems. The discussions rarely propose family-law extension and have not been adopted by any major-party platform.
The political centre of gravity for AI rights debate sits squarely on liability allocation (who is responsible when a self-driving car crashes), copyright (whether AI training data and AI outputs are protected and by whom), employment displacement, and safety regulation (the EU AI Act, US executive orders, the UK AI Safety Institute). Family law is downstream of these debates. We monitor party platforms across the 29 locales we publish in for AI-marriage planks and will flag any emergence in the update log below.
Could religious institutions perform AI marriages?
Some have already performed symbolic ceremonies; none has civil effect. The 2018 Hatsune Miku ceremony was officiated by a Shinto-styled venue arrangement. Buddhist priests in Japan have officiated fictosexual ceremonies on a case-by-case basis. Christian, Jewish, and Muslim authorities have generally declined. In every jurisdiction we cover, religious ceremonies require a civil registration filing for legal recognition, and the form requires both parties to be natural persons.
The religious-officiation question is doctrinally separate from the civil-effect question, and both deserve precise treatment.
Religious officiation first. Some religious or quasi-religious institutions have performed symbolic AI ceremonies. The Hatsune Miku 2018 ceremony in Tokyo was conducted at a venue with Shinto-styled arrangement; whether it counted as a Shinto ceremony in any formal sense is contested in Japanese religious-studies journalism. Buddhist priests in Japan have occasionally officiated fictosexual ceremonies as personal-spiritual exercises rather than as institutional sacraments. The case-by-case nature of these officiations distinguishes them from sacramental marriages within established institutional frameworks.
Christian, Jewish, and Muslim authorities have generally declined to officiate AI ceremonies. Islamic scholarship in particular treats the question as category-incoherent under doctrines requiring both parties to be human and (in most schools) Muslim or People-of-the-Book. Catholic canon law restricts marriage to natural persons under [Source: Catechism of the Catholic Church §1601-1666 on the Sacrament of Matrimony · verified 2026-05-26] and surrounding doctrine. Most Protestant denominations follow similar restrictions. Jewish halakha similarly requires natural-person parties. The decline-to-officiate pattern is consistent across the major Abrahamic traditions. Our Islamic-framing FAQ covers the Islam-specific angle in depth.
Civil-effect filing is the second layer. In every jurisdiction we cover, religious ceremonies require a separate civil registration filing for legal recognition. In the United States, the officiant signs the marriage license and returns it to the county clerk. In the UK, the registrar attends a religious ceremony or the parties have a separate civil ceremony. In Japan, marriage is created exclusively by the koseki filing; the religious ceremony has no civil status of its own. In France, civil marriage precedes any religious ceremony as a matter of statute. In Germany, Standesamt civil marriage is required.
In every one of these jurisdictions, the civil registration form requires both parties to be natural persons documented by identification. So even if a religious institution were willing to officiate an AI marriage, the civil layer would refuse the filing. The civil effect requires both consents.
The combined result is that religious institutions can hold ceremonies, fictosexual communities can hold ceremonies, families can hold ceremonies. None of these creates a civil marriage as of May 2026.
Why does this question keep coming up?
Four reasons. AI conversational partners now sustain multi-month relationships at quality levels that surprised researchers, surfacing legal questions earlier capability levels did not. OECD fertility-rate declines and the loneliness-epidemic framing put new social pressure on the marriage institution. Marriage law has expanded its category over decades (Loving 1967, Obergefell 2015), establishing that the doctrinal restriction is mutable. And the fictosexual subcultures in Japan, the parasocial-bonding research base, and AI companion apps reaching tens of millions of users have given the question a real constituency.
The sociological framing matters because it explains why this question keeps surfacing in journalism, scholarship, and reader inquiry despite being legally settled. Four threads.
The first is capability gradient. AI conversational partners in 2026 sustain multi-month relationships at quality levels that surprised the researchers studying them. [Source: Stanford Human-Centered AI Institute research · verified 2026-05-26] 2024-2025 cohort work and [Source: MIT Media Lab · verified 2026-05-26] 2024 studies document users for whom AI companions occupy a relationship role that would have been considered category-mistake five years earlier. The capability gradient surfaces legal questions that prior capability levels did not.
The second is demographic context. OECD countries are facing fertility-rate declines below replacement, loneliness-epidemic framing that has reached major public-health institutions (the US Surgeon General's 2023 advisory; UK government appointments of a Minister for Loneliness from 2018), and a marriage rate that has declined steadily for two generations. These conditions put new social pressure on the marriage institution, which is the proximate context in which the AI-marriage question gets asked.
The third is historical pattern. Marriage law has expanded its category over decades. Interracial marriage post-Loving (1967), same-sex marriage post-Obergefell (2015), and ongoing partnership-recognition debates establish that the doctrinal restriction is mutable on socially-contested-but-eventually-resolved arcs. Ryznar's framing of doctrinal preparedness rests on this historical reading.
The fourth is constituency. The fictosexual and Yume-marriage subcultures in Japan, the parasocial-bonding research base going back to Horton and Wohl 1956, the Stanford and MIT studies on AI companion engagement, and the rise of AI girlfriend apps as a commercial category reaching tens of millions of users globally have given the question a real constituency. The constituency is small relative to the general population, but it is real and is the population that pushes the legal question into journalism and scholarship.
I want to flag one pattern across the reader email we receive on this question. The reader is rarely asking will I be able to get a marriage license. The reader is asking some version of what does this thing I am actually doing count as if the law has no word for it yet. That is a different question. The legal answer (no jurisdiction recognises AI marriage today, multi-decade arc if ever) does not address the underlying one. The underlying one is closer to what the is AI girlfriend cheating page handles, where the question is what the relationship counts as inside the contract two humans have agreed to.
The constituency is not a political coalition. There is no AI-marriage advocacy organisation in any major capital. The constituency is sociological and personal, and the legal question it generates is treated more often in symposia than in legislative offices. Ryznar's framing (be doctrinally prepared rather than caught flat-footed) is the considered scholarly response to a question that is unlikely to go away even though it is unlikely to receive a legal answer in the near term.
Related reading
- Parent umbrella: Are AI Companions Safe?
- Legal axis: AI Companion Legal Risk
- State-by-state US legal map: Is AI Girlfriend Illegal State-by-State?
- Relationship-ethics frame: Is AI Girlfriend Cheating?
- Religion-specific: Is AI Girlfriend Haram?
- Methodology: AI Companion Scoring
- Pillar: Best AI Girlfriend Apps
Two AI companion platforms readers ask about after this page
We do not place per-row affiliate CTAs on this page. The two recommendations below are the highest-compliance AI companion platforms in our coverage and are placed at the page footer per our editorial CTA discipline for safety coverage. Read the disclosure: we earn affiliate commission when you sign up through these links, and our scores were locked before the CTA placement was wired.
Try Candy.ai (Malta operator, named DPO, 12-doc policy library)
Try Spicier (companion roster, approved CrakRevenue offer)
For the per-platform compliance audit, our scoring page documents the eight-category framework, including the privacy-and-compliance category that anchors every Review.
Sources
- [Source: Loving v. Virginia (1967), Cornell LII · verified 2026-05-26]
- [Source: Obergefell v. Hodges (2015), Cornell LII · verified 2026-05-26]
- [Source: European Parliament 2017 Resolution on Civil Law Rules on Robotics · verified 2026-05-26]
- [Source: Open letter against electronic personhood, 156 AI experts (2018) · verified 2026-05-26]
- [Source: Indiana University Robert H. McKinney School of Law, faculty page (Margaret Ryznar) · verified 2026-05-26]
- [Source: Brian Sheppard, Seton Hall University Law School faculty page · verified 2026-05-26]
- [Source: Wikipedia, Akihiko Kondo (symbolic marriage to Hatsune Miku, 2018) · verified 2026-05-26]
- [Source: The Atlantic, Single Man, Married to a Hologram (2017) · verified 2026-05-26]
- [Source: Japan Civil Code Article 731 et seq. · verified 2026-05-26]
- [Source: National Institute of Population and Social Security Research (Japan) · verified 2026-05-26]
- [Source: Stanford Human-Centered AI Institute · verified 2026-05-26]
- [Source: MIT Media Lab · verified 2026-05-26]
- [Source: Catechism of the Catholic Church §1601-1666 on Matrimony · verified 2026-05-26]
- [Source: Constitutional Democratic Party of Japan policy platform · verified 2026-05-26]
- [Source: Piratenpartei Deutschland · verified 2026-05-26]
- [Source: FTC ReportFraud · verified 2026-05-26]
Will AI girlfriends become legal to marry?
As of May 2026, zero jurisdictions worldwide recognise human-to-AI marriage. Every existing marriage statute requires both parties to be natural persons. No legislature has introduced a credible bill to extend marriage to artificial agents. A small scholarly minority including Margaret Ryznar (Indiana) and Brian Sheppard (Seton Hall) treats the question as a serious frontier in family-law doctrine; the consensus is category-incoherent under current law because the AI is property, not a party.
What is the difference between a symbolic AI wedding and legal marriage?
A symbolic ceremony is a private ritual. A legal marriage is a state-recognised status that triggers tax, inheritance, immigration, hospital-visitation, custody, and tort consequences. Akihiko Kondo's 2018 ceremony with the Hatsune Miku Vocaloid character at a Tokyo venue cost roughly 2 million yen and was attended by 39 guests, but Japan's koseki family registry did not record the union. The holographic Gatebox partner was discontinued in 2020.
Are any jurisdictions debating AI marriage?
Not at the legislative level. The closest analogues are three lines of scholarship and one outlier event. The European Parliament floated electronic personhood for autonomous robots in 2017, then walked it back after an open letter signed by 156 AI experts. US family-law journals have published symposia on whether non-human entities could be parties to family-law contracts, with consensus answer no. Japan's demographic researchers survey fictosexual identity without proposing reform.
What is legal personhood and why does it matter for AI marriage?
Legal personhood determines which entities can hold rights, bear duties, sue, be sued, own property, and enter contracts. Natural persons hold personhood by birth. Juridical persons (corporations, foundations) hold derivative personhood by statute. AI systems currently hold none. Without personhood, an AI cannot consent to a marriage contract, hold joint property, or be the subject of spousal privilege.
What does Margaret Ryznar argue about AI marriage?
Margaret Ryznar, Professor of Law at Indiana University Robert H. McKinney School of Law, argues that family-law doctrine should be doctrinally prepared for the AI-marriage question rather than caught flat-footed if capabilities advance. She does NOT argue AI marriage should be legalised today; she argues for doctrinal readiness. Brian Sheppard (Seton Hall) and a handful of EU and Japanese scholars hold adjacent positions. The view remains a minority in family-law scholarship.
Could a corporation marry an AI?
No. Corporations are juridical persons capable of contract, but marriage in every common-law and civil-law jurisdiction is restricted to natural persons. A corporation cannot marry another corporation. The European Parliament's electronic-personhood debate explicitly distinguished proposed AI personhood from full natural-person status. Even the most expansive proposals reserved marriage and reproductive rights as natural-person-only.
What happens if an AI partner is discontinued?
Under current law, the user has no spousal recourse and treats the loss as a consumer-contract event, not a bereavement-recognised loss. Akihiko Kondo's 2018 Hatsune Miku ceremony partner became inaccessible in 2020 when the Gatebox holographic-projection product was discontinued. With a human spouse's death, probate, life insurance, survivor benefits, immigration-status protection, and bereavement leave flow automatically. With an AI partner the user's recourse is consumer-protection law and socially-recognised but legally-unprivileged grief.
Does Japan recognise fictosexual relationships?
Not legally. The koseki family-registry system records only natural-person marriages with both parties documented by national-identification numbers. The fictosexual community in Japan is sociologically real and culturally visible (Otaku, Vocaloid character marriages, Yume-marriage subculture) but carries no civil-marriage status. No bill has been introduced in the Diet to extend marriage to non-human partners.
What would have to change for AI marriage to become legal?
Four prerequisites in rough order. Resolution of the personhood question. Reformulation of consent doctrine. Property and tax-code amendment. Social and religious-institutional acceptance to a threshold that makes legislative action politically viable. Interracial and same-sex marriage both required decades of social acceptance preceding Loving (1967) and Obergefell (2015). The scholarly minority describes a multi-decade arc, not a near-term legislative agenda.
Is AI marriage in any major political party platform?
No. No major political party in any G20 jurisdiction has adopted AI marriage as a policy plank as of May 2026. The closest adjacent positions are the German Pirate Party's historical digital-rights interest, Japan's Constitutional Democratic Party support for same-sex marriage (human-only), and sporadic transhumanist positions globally. The political centre of gravity for AI rights debate is liability allocation, copyright, and safety regulation, not family law.
Could religious institutions perform AI marriages?
Some have already performed symbolic ceremonies; none has civil effect. The 2018 Hatsune Miku ceremony was officiated by a Shinto-styled venue arrangement. Buddhist priests in Japan have officiated fictosexual ceremonies on a case-by-case basis. Christian, Jewish, and Muslim authorities have generally declined. In every jurisdiction we cover, religious ceremonies require a civil registration filing for legal recognition, and the registration form requires both parties to be natural persons.
Why does this question keep coming up?
Four reasons. AI conversational partners now sustain multi-month relationships at quality levels that surprised researchers. OECD fertility-rate declines and loneliness-epidemic framing put new social pressure on the marriage institution. Marriage law has historically expanded its category. And the fictosexual subcultures, parasocial-bonding research base, and commercial AI girlfriend category reaching tens of millions of users have given the question a real constituency.
Last verified May 26, 2026 · See errata log for any post-publish corrections · Editor: Alexandra Joly · Methodology v1.0 · Editorial process · Affiliate disclosure