Ethics of Organoid Intelligence: Consent & Consciousness

The ethics of organoid intelligence covers the moral questions raised by computing with lab-grown human brain tissue: whether cell donors truly consented, how their genomic privacy is protected, whether organoids could ever be conscious, how animal transplants are governed, and who benefits commercially. Today’s organoids are far from brains, but the field is building its ethical framework now.

Key takeaways

Most brain organoids come from induced pluripotent stem cell (iPSC) lines, and many of those lines were collected under consent forms that never anticipated biocomputing.
There is no evidence that current organoids are conscious. A 2021 National Academies report called awareness, emotion or pain “extremely unlikely in the foreseeable future,” while urging that the question be revisited as complexity grows.
The Baltimore declaration (Hartung et al., 2023) and the organoid intelligence roadmap (Smirnova et al., 2023) commit the field to embedded ethics and public engagement from the start.
Donor genomes are identifiable information. Labs need access controls, clear data policies and honest consent language.
Responsible communication matters: overstating what organoids do damages public trust as much as understating the ethical stakes.

What is organoid intelligence, and why does it raise ethical questions?

Organoid intelligence (OI) is the research program of using three-dimensional cultures of human neural tissue as computing substrates. The tissue is a brain organoid: a self-organizing cluster of neurons and supporting cells grown from stem cells, first described in its modern “cerebral organoid” form by Lancaster et al. in Nature (2013). Researchers place these organoids on a microelectrode array, record their action potentials, stimulate them, and study whether their activity can be shaped to perform useful work. For a broader introduction, see our guide to what organoid intelligence is.

Three features make this ethically distinctive compared with ordinary cell culture or silicon biocomputing research:

The material is human and traceable to a person. Every organoid carries the full genome of a living or deceased donor.
The tissue is neural. It is the kind of tissue that, in a body, underlies experience. That invites questions about moral status that skin or liver organoids do not.
The goal is function, including learning. OI research deliberately tries to make organoids more capable, through closed-loop stimulation and training, which is exactly the direction that would matter if moral status were ever at stake.

None of this means current work is wrong. It means the field has a duty to think ahead, which is the central argument of both the Baltimore declaration and the OI roadmap.

From skin or blood to neurons

Almost all brain organoids used in OI start as induced pluripotent stem cells. Takahashi and Yamanaka showed in Cell (2006) that adult mouse cells could be reprogrammed to a pluripotent state with a small set of factors, and Takahashi et al. (2007) did the same with adult human fibroblasts. That breakthrough means an organoid can be grown from a skin biopsy or blood draw, with no embryo involved. The process is described step by step in our article on how brain organoids are grown.

Avoiding embryos removed one major ethical controversy, but it did not remove the donor. An iPSC line is, genetically, that person. It can be expanded almost indefinitely, shared between labs, banked, and differentiated into many tissue types, including neural tissue that is then trained on computational tasks.

The problem is that consent for many existing lines was obtained years ago, for broad “research use,” often before brain organoids or biocomputing existed. The National Academies of Sciences, Engineering, and Medicine addressed this directly in its 2021 report The Emerging Field of Human Neural Organoids, Transplants, and Chimeras. It found that specific consent for organoid, transplant or chimera research was generally not obtained for past materials, that recontacting donors is sometimes impossible, and that some donors might have objected had they been asked. It also noted that past ethics violations involving African American and Native American research participants make this issue especially sensitive.

Under US rules, deidentified materials collected with appropriate consent can still be used, but the report notes that some view this as a minimal standard. The committee did not reach a firm conclusion on whether specific consent should be required for new collections, and the debate remains open.

Provenance as an ethical tool

The practical response is provenance: knowing, for each cell line, who collected it, under what consent form, with what permitted uses and what restrictions. Provenance turns an abstract ethical concern into something a lab can check before an experiment starts. Key questions include:

Does the consent explicitly or plausibly cover neural differentiation and computational use?
Does it allow commercial use, and was the donor told?
Can the donor withdraw, and what happens to derived lines and data if they do?
Did the donor agree to genomic data sharing, and at what level of access?
Does the line come from a repository with documented governance?

The International Society for Stem Cell Research (ISSCR) Guidelines for Stem Cell Research and Clinical Translation, updated in 2021 and summarized by Lovell-Badge et al. in Stem Cell Reports, set expectations around informed consent for donating materials and around oversight of different categories of stem cell research. They are the most widely referenced international guidance for this field, though they are professional guidelines rather than law.

Could a brain organoid be conscious?

This is the question that draws the most public attention, and it deserves a careful answer rather than either alarm or dismissal.

What the evidence says today

There is no evidence that any brain organoid made to date is conscious, sentient in the everyday sense, or able to feel pain. The National Academies report concluded that, given their limited complexity, organoids having awareness, emotion or the experience of pain is “extremely unlikely in the foreseeable future,” and that they currently have no more moral standing than other human neural tissue in a dish. It also found that measurements of neuronal activity alone are not sufficient to show consciousness or pain, and that most existing assessment methods rely on whole-animal behavior, which organoids do not have.

The OI roadmap is equally explicit. Smirnova et al. (2023) state that the program does not aim to recreate human consciousness, while acknowledging that concerns about consciousness and pain are legitimate and need ongoing attention.

Why current organoids are very far from brains

The gap between an organoid and a brain is not a matter of degree on one axis. It is a gap on almost every axis that plausibly matters for experience.

Property	Human brain	Typical brain organoid today
Neuron count	About 86 billion neurons (Azevedo et al., 2009)	Fewer than 100,000 cells for many organoids (Smirnova et al., 2023)
Size	A whole organ filling the skull	Often below 500 micrometers in diameter (Smirnova et al., 2023)
Blood supply	Dense vascular network delivering oxygen and nutrients	No vasculature; relies on diffusion, which can only supply cells within a few hundred micrometers of the surface, so the core starves
Sensory input	Continuous input from eyes, ears, skin, body	None, apart from electrical or chemical stimulation chosen by researchers
Motor output	Controls a body that acts in the world	None, unless coupled to software or a device
Organization	Layered cortex, long-range connections between dozens of regions, thalamic loops	Partial self-organization; regional identity and layering only roughly resemble development
Developmental stage	Keeps maturing long after birth	Even long-term cultures resemble prenatal or early postnatal stages (Gordon et al., 2021)
Cell health	Regulated environment	Cells show elevated stress responses that affect cell identity (Bhaduri et al., 2020)

Lack of organoid vascularization is a good example of why this matters. Without blood vessels, cells deep inside an organoid receive too little oxygen and nutrients, so larger organoids tend to develop a dying core, which caps their size and complexity. Bhaduri et al. reported in Nature (2020) that cortical organoids show activated cellular stress pathways that impair the specification of neuronal subtypes, a further sign that organoid tissue is not simply a small brain.

Organization matters too. Many leading theories of consciousness emphasize large-scale integration, recurrent connectivity between regions, and interaction with a body and environment. Organoids have local networks that produce network bursts and oscillations, but they lack the architecture those theories point to.

Why the question still matters

Three things keep the question alive. First, organoids are improving: researchers are working on vascularization, longer culture, and assembloids that join several brain-region organoids together. Second, some organoid activity looks superficially brain-like. Trujillo et al. reported in Cell Stem Cell (2019) that cortical organoids develop complex oscillatory activity, and that some features of this network activity resembled patterns in EEG recordings from preterm infants. That is a real and interesting result about network development, but it is not evidence of experience. Third, we lack an agreed test for consciousness even in humans with brain injury, let alone in tissue with no behavior.

Lavazza and Massimini, writing in the Journal of Medical Ethics (2018), argued that consciousness assessment methods developed for patients might eventually be adapted to organoids, and that the field should prepare such tools before they are needed. Farahany et al. made a related case in Nature (2018), calling for ethical guidance as brain surrogates become more sophisticated. The sensible stance is precaution proportional to evidence: no panic about today’s organoids, and agreed checkpoints for tomorrow’s.

How should we talk about organoid intelligence without hype?

Language shapes public trust, regulation and funding. The field has already had one instructive episode. Kagan et al. reported in Neuron (2022) that cultured neurons on a high-density array learned to play a simplified version of Pong in a closed loop, a study often called DishBrain. The paper’s title used the word “sentience,” which the authors defined in a specific, technical sense linked to the free energy principle. A group of neuroscientists responded in the same journal (Balci et al., 2023), arguing that terms like “sentience” and “intelligence” in that context risked misleading readers. Whatever one’s view of that exchange, it shows how a single word can travel far beyond its technical meaning. Our article on how neurons learn in a dish covers the experiment itself.

Practical rules for researchers, companies and journalists:

Prefer “brain organoid” or “neural organoid” over “mini-brain.” The nomenclature reflects what the tissue is.
Report what was measured. Spikes, evoked responses, changes in firing after stimulation, task scores. Say what was not measured.
Give scale. “Fewer than 100,000 cells, compared with about 86 billion neurons in a human brain” is more informative than “brain-like.”
Separate adaptation from cognition. A network whose responses change with feedback is showing synaptic plasticity and adaptation; that is not the same as thinking.
Avoid emotional verbs such as “wants,” “feels” or “suffers” unless a study tested for something relevant.
State limitations alongside results, as good papers do.

Hype is not only a communication problem. It is an ethical one, because it can mislead donors about what their cells are being used for, mislead investors, and provoke public backlash that harms legitimate research.

What do the Baltimore declaration and the OI roadmap commit to?

The Baltimore declaration

The Baltimore declaration toward the exploration of organoid intelligence was published by Hartung et al. in Frontiers in Science in February 2023. It came out of the first Organoid Intelligence workshop, held online in February 2022. The declaration calls on the international community to pursue OI for understanding the brain and for new forms of biocomputing, while recognizing that its ethical challenges are largely unexplored. It asks for ongoing, stringent discussion with all relevant stakeholders, aimed at an ethical framework that takes public values into account. It names the possibility that organoids could develop aspects of consciousness as something to watch for and guard against, and it names the personal rights and interests of cell donors as an important consideration.

Embedded ethics in the roadmap

The companion roadmap by Smirnova et al. (2023) describes how this should work in practice. It proposes an embedded ethics approach, in which ethicists work alongside research teams continuously, identifying and analyzing issues as they arise rather than signing off once at the start. The authors note that embedded ethics has no built-in way to capture public values, so they add a three-way feedback loop between researchers, ethicists and the public, including deliberative engagement with diverse groups such as neurodiversity advocates.

The roadmap also raises specific donor questions: privacy, whether donors should be told about health-relevant findings that emerge from their cells, and whether donor rights extend beyond the moment of donation. And it discusses what bioethicists call the Greely dilemma: the more successfully organoids model the human brain, the more they may raise the very concerns that made human tissue worth studying in the first place.

What guidance already applies?

OI does not sit in a regulatory vacuum. It inherits rules from stem cell research, human tissue research and data protection. The table below summarizes the main reference points we were able to verify.

Source	Type	What it addresses
ISSCR Guidelines for Stem Cell Research and Clinical Translation (2021 update)	International professional guidelines	Informed consent for donated materials, tiered oversight of stem cell research, including chimera experiments
National Academies report on neural organoids, transplants and chimeras (2021)	US consensus study	Moral status, consciousness assessment, consent for past and future materials, chimera welfare and oversight
Baltimore declaration (Hartung et al., 2023)	Community statement	Commitment to explore OI with stringent ethical discussion, consciousness vigilance, donor rights
OI roadmap (Smirnova et al., 2023)	Research program paper	Embedded ethics, public engagement, donor privacy and incidental findings
Institutional review and data protection law	Local and national law	Human subjects protections, genomic data handling (varies by jurisdiction)

Labs should also follow their own institution’s human research ethics processes and local data protection law, which vary by country and are outside the scope of this article.

What about transplanting organoids into animals?

Some of the most striking organoid studies involve transplantation. Mansour et al. reported in Nature Biotechnology (2018) that human brain organoids transplanted into the mouse brain became vascularized by the host and showed functional activity. Revah et al. reported in Nature (2022) that human cortical organoids transplanted into the somatosensory cortex of newborn rats matured further than in a dish, received inputs from the rat’s thalamus and cortex, responded to sensory stimulation, and could drive reward-seeking behavior when activated optogenetically.

These studies show how much richer organoid tissue becomes when it gets blood supply and real inputs, which reinforces the point that isolated organoids are developmentally limited. They also raise their own questions about the welfare of host animals and about how “humanized” an animal brain should be allowed to become. Chen et al. reviewed this in Cell Stem Cell (2019), and the National Academies report found that host animals should be monitored for distress-like behaviors, and that more complex chimeras and primate work may need national-level ethical discussion beyond local animal care committees.

For OI specifically, the important point is that biocomputing on high-density microelectrode arrays does not require animal transplantation. Some in the field present organoid systems as a potential route to reducing animal use in research, though whether that happens depends on validation work that is still in progress.

How should donor genomes and data be protected?

An organoid experiment produces two kinds of sensitive material: the cell line itself, which contains the donor’s genome, and data, including any genomic sequencing and the electrophysiology recorded from the tissue.

Genomic data is identifiable even when names are removed. Gymrek et al. showed in Science (2013) that researchers could infer the surnames of supposedly anonymous research participants by matching markers on the Y chromosome to recreational genealogy databases. Genetic information is also shared with relatives, so a donor’s data can reveal facts about people who never consented to anything.

Electrophysiology recordings on their own are much less sensitive, but they become linked to the donor once they are combined with cell line identifiers and metadata. Good practice is to:

Store cell line identifiers separately from any donor-identifying information.
Use controlled-access repositories for genomic data rather than open publication, unless the donor explicitly agreed to open release.
Share electrophysiology openly where consent allows, using standards like Neurodata Without Borders with metadata fields that record consent scope rather than donor identity.
Write a policy for incidental findings, such as a disease-associated variant discovered while characterizing a line, before it happens.

Open data and privacy are not opposites. Clear consent and good metadata make it possible to share electrophysiology widely while keeping genomic information under control.

Who should benefit from commercial organoid computing?

If organoid systems become commercially valuable, value will flow from tissue that originated with donors. History offers a cautionary tale. Cells taken from Henrietta Lacks in 1951 became the HeLa line, one of the most widely used in biology, without her knowledge. In 2013 the US National Institutes of Health reached an agreement with her family on controlled access to HeLa genome data, with family members involved in reviewing requests, as reported by Nature. It is a widely cited example of donor involvement added decades too late.

There is no settled model of benefit sharing for biocomputing. Options discussed in bioethics more broadly include:

Transparent consent that says plainly that commercial use may occur and that donors will not receive royalties, if that is the case.
Community benefit, such as support for patient groups connected to disease-model lines.
Open science commitments, including open data and protocols, so public benefit does not depend on a single company.
Governance roles for donor or community representatives in decisions about sensitive uses.

Commercialization also creates conflicts of interest in how results are described. A company has incentives to describe its systems as more brain-like or more capable than evidence supports. That is another reason the communication rules above matter.

What does a responsible organoid intelligence lab look like?

Here is a practical checklist drawn from the sources above. It is not legal advice and does not replace institutional review, but it turns principles into actions.

Document provenance for every cell line: source repository, consent form version, permitted uses, commercial terms and withdrawal procedures.
Prefer lines with consent that covers neural differentiation, data sharing and, if relevant, commercial use. Where consent is ambiguous, seek ethics review before use.
Record consent scope as machine-readable metadata attached to every experiment and dataset, so downstream users inherit the restrictions.
Separate identity from data: keep donor-identifying information out of experimental records and use controlled access for genomic data.
Write an incidental findings policy before characterizing lines genetically.
Embed ethics: involve an ethicist or ethics advisory group in project planning, not only at approval stage, following the OI roadmap’s embedded ethics model.
Set complexity checkpoints: agree in advance which changes (vascularization, much larger cell counts, sensory-like inputs, long-duration training, multi-region assembloids) trigger fresh ethical review.
Avoid animal transplantation unless the research question requires it, and follow chimera oversight rules if it does.
Communicate precisely: describe measurements, scale and limitations, and avoid language that implies experience.
Publish openly where consent allows, using standard formats, so claims can be checked and duplicated work avoided.
Engage the public, including groups whose cells or conditions are being studied.

A table is a useful way to see how these map to specific risks.

Risk	Mitigation from the checklist
Use beyond what donors agreed to	Provenance records, consent scope metadata, ethics review for ambiguous lines
Re-identification of donors	Separate identity from data, controlled genomic access
Drift toward morally significant complexity	Embedded ethics, complexity checkpoints
Animal welfare concerns	Avoid transplants by default, follow chimera oversight
Public mistrust from hype	Precise communication, open data
Unfair distribution of benefits	Transparent consent, community benefit, governance roles

Where can you go deeper on each question?

This guide is the overview. Each of the hardest questions has its own in-depth essay:

Consciousness. Can brain organoids become conscious? compares what the leading theories of consciousness predict for organoids and why proposed tests are so hard to apply.
Moral status. Do biological computers have moral status? sets out the grounds philosophers use, and what each would mean for living computers.
Ownership and consent. Who owns a living computer? follows the chain from donor to cell line to computing service, with the lessons of HeLa.
End of life and welfare. Is it wrong to switch off a biological computer? looks at pain, welfare under uncertainty and responsible retirement policies.

What should the field do next?

The ethics of organoid intelligence is not a reason to stop the research. It is a set of obligations that grow as the science grows. Today’s organoids are small, avascular, developmentally immature, and cut off from sensory input; there is no good reason to think they experience anything. But the explicit goal of OI is to make them more capable, and the field’s own founding documents say that ethical vigilance must keep pace. As the future of biocomputing brings larger, longer-lived and better-connected organoids, the checkpoints in this article will matter more, not less.

The most useful near-term work is unglamorous: consent forms that actually describe biocomputing, provenance records attached to every dataset, ethicists embedded in research teams, and honest descriptions of results. Building that infrastructure into the tools researchers use every day is part of what we are working on at Organoid Programming, where provenance and consent metadata are part of the platform design rather than an afterthought.

Frequently asked questions

Are brain organoids conscious?

There is no evidence that any brain organoid made so far is conscious. A 2021 National Academies report judged awareness, emotion or pain extremely unlikely in the foreseeable future, given how small, unvascularized and disconnected from sensory input today’s organoids are. The same report said the question should be revisited as organoids become more complex, which is why researchers argue for ongoing monitoring rather than a one-time answer.

Do cell donors have to consent to organoid intelligence research?

Most organoids come from induced pluripotent stem cell lines made from donated skin or blood. Many older lines were collected under broad research consent that never mentioned organoids or computing. Ethics bodies, including the National Academies, have noted that specific consent was generally not obtained and that this is debated. Responsible labs check each line’s consent scope and prefer lines whose donors agreed to uses like these.

What is the Baltimore declaration on organoid intelligence?

The Baltimore declaration is a 2023 statement in Frontiers in Science by participants of the first Organoid Intelligence workshop, held online in February 2022. It calls on scientists to explore organoid intelligence for understanding the brain and for biocomputing, while treating its ethical questions seriously, including the possibility of consciousness and the rights and interests of cell donors.

What is embedded ethics in organoid intelligence?

Embedded ethics means ethicists work inside research programs from the start rather than reviewing them afterwards. The organoid intelligence roadmap by Smirnova and colleagues (2023) proposes this approach and adds a feedback loop with the public, so researchers, ethicists and wider communities shape decisions together as the science develops and new questions appear.

Why is donor genome privacy an issue for biocomputing?

Every cell in an organoid carries the donor’s genome, and sequencing data or the cell line itself can reveal information about the donor and their relatives. Research has shown that supposedly anonymous genomes can sometimes be re-identified, for example by matching Y-chromosome markers to genealogy databases. Labs should treat genomic data as identifiable, control access and avoid publishing raw sequences without consent.

Is putting human brain organoids into animals allowed?

Transplanting human brain organoids into animals is legal in many jurisdictions and has been done in mice and rats, but it is treated as sensitive. Guidance such as the ISSCR Guidelines calls for specialized review of chimera research, and the National Academies said more complex chimeras may need national-level discussion. Biocomputing on microelectrode arrays does not require animal transplantation.

How should companies share benefits with cell donors?

There is no settled rule. Options discussed in bioethics include clear disclosure at consent that commercial use may happen, community or patient-group benefits, open data, and governance roles for donor representatives. The 2013 agreement between the NIH and the family of Henrietta Lacks over HeLa genome data is often cited as an example of retrofitting donor involvement after the fact.

How can journalists describe organoid intelligence without hype?

Use precise terms: organoids are small clusters of lab-grown neural tissue, not mini-brains with thoughts. Report what was measured (spikes, stimulus responses, task performance) rather than what it might mean. Give scale comparisons, name the limitations the authors state, and avoid words like ‘thinking’ or ‘feeling’ unless a study actually tested for them.

The ethics of organoid intelligence: consent, consciousness and responsible research