Designer Babies: The Reality and Myths of Genetic Optimization in Embryos

Several companies will now screen an IVF embryo against a polygenic score and hand prospective parents a ranked list — this one lower risk for heart disease, that one predicted taller or brighter. The service is real, it is commercially available, and people are paying for it.

So we searched the public trial registry for the evidence behind it. On 9 October 2026, ClinicalTrials.gov held exactly two registered studies of polygenic embryo selection. Both are sponsored by companies that sell the test. Neither is randomised, and neither measures whether the resulting children differ in the traits or diseases the scores claim to predict.

For comparison, the same registry lists 220 trials of antisense oligonucleotides and 2,933 involving CAR-T therapy. Genetic medicine is not short of clinical evidence. This particular application has generated almost none — and it is being sold to people making the most consequential decision of their lives.

In June 2025, a genomics company called Nucleus Genomics launched a product with a science-fiction premise and a real-world price tag. For around $5,999, Nucleus Embryo promised to help parents rank their IVF embryos by predicted intelligence, longevity, disease resistance, body mass index, eye colour and even left-handedness.

The pitch was control over a child’s biological future. The reality, as the peer-reviewed evidence shows, is considerably smaller than that promise implies.

The dream of choosing a child’s traits is old, running from Plato’s Republic to Huxley’s Brave New World to the film GATTACA. What is new is a company selling it as a subscription.

The gap between what the marketing implies and what the peer-reviewed science actually supports is where this whole story lives, and it is wider than the price tag suggests.

This article separates reality from hype. It explains what embryo screening genuinely does, why predicting complex traits from embryo DNA is far weaker than advertised, how small the real gains are once you read the studies, and why the ethics matter regardless.

What Preimplantation Genetic Testing Actually Does

The technology at the core of embryo screening is preimplantation genetic testing, or PGT. It is a well-established part of IVF, used for decades to check embryos for chromosomal errors and specific inherited mutations before one is transferred to the womb.

PGT works by taking a tiny biopsy from an embryo at the blastocyst stage, around five to six days after fertilisation, when a handful of cells can be removed from the outer layer that will go on to form the placenta. The results show which embryos do not carry a targeted mutation.

In its most established form, testing for monogenic conditions (PGT-M), the technology is genuinely powerful. It flags embryos carrying single-gene mutations with clear effects: cystic fibrosis, sickle cell disease, Huntington’s disease, Tay–Sachs, and the BRCA1 and BRCA2 cancer-risk mutations.

For a family with a known inherited condition, PGT-M offers something profound and legitimate: the chance of a biological child free from that specific disease. The UK regulator has approved more than 2,000 conditions for this kind of testing.

This connects directly to the wider science of inherited disease covered in our guide to the genetics of cancer.

Testing for aneuploidies (PGT-A) instead screens for the wrong number of chromosomes, a major cause of failed implantation and miscarriage. Its clinical value, however, is genuinely contested rather than settled.

The distinction matters. PGT-M answers a precise yes-or-no question about a single well-understood gene. Trait prediction asks something far harder: to forecast an outcome shaped by thousands of variants and a lifetime of environment. Only the first is on solid clinical ground.

Large randomised trials have muddied the picture. A 2021 trial in the New England Journal of Medicine, run across fertility centres in China, randomly assigned 1,212 women with at least three good-quality embryos to PGT-A or to conventional IVF. Live births followed for 77.2% of the women in the PGT-A group and 81.8% in the conventional group. For good-prognosis patients, the screening did not help.

What none of these methods can do is reliably predict complex traits such as intelligence, personality or athleticism, which depend on interactions between thousands of genes and a lifetime of environment. That gap is where the marketing outruns the science.

The Problem With Polygenic Scores

Nucleus Embryo and similar services rest on polygenic scores: statistical summaries of the combined effect of thousands of genetic variants, each with a minuscule individual influence, on a given trait. At the population level these scores have real predictive value.

The catch is that population-level validity does not transfer cleanly to a single embryo. Four specific limitations make polygenic scores scientifically problematic for embryo selection, and each one narrows the promised benefit considerably.

The effect sizes are small

A polygenic score for educational attainment explains only about 12 to 16% of the variance in years of education, and cognitive scores capture less still. Within families, where embryo selection actually operates, the usable signal is roughly half that again.

Scores travel poorly across ancestries

Most large genetic studies were done in people of European descent, so the scores lose accuracy elsewhere. Research has found their predictive power falling by roughly two-and-a-half-fold in East Asian populations and nearly five-fold in those of African ancestry.

This is a serious equity problem baked into the method. One analysis of the first decade of polygenic-scoring studies found that around two-thirds used exclusively European-ancestry participants, meaning the tools work best for exactly the populations already best served by medicine.

Complex traits really are complex

Intelligence is not a simple sum of independent variants. It involves gene–gene interactions, gene–environment feedback and epigenetic effects that no current score captures — a theme explored in our piece on how the environment shapes gene expression.

A score predicts averages, not individuals

The embryo with the highest score is not guaranteed to become the child with the most of that trait. It is merely the one that, averaged across many similar embryos, would be expected to score slightly higher.

There is a subtler trap too. Siblings share much of their DNA, so the differences a score can act on within one family are far smaller than those it detects across a whole population. That is why real gains shrink below the headline correlations.

How Small the Real Gains Are

Peer-reviewed work has actually measured the likely benefit, and the figures are sobering. A 2019 study in the journal Cell by Ehud Karavani and colleagues modelled selecting the top-scoring embryo from a realistic IVF batch.

Their conclusion was that the average gain would be around 2.5 centimetres of height or roughly 2.5 IQ points. Crucially, they added that these averages come with wide prediction intervals, and that in large families the child with the highest score is usually not the tallest.

A 2021 analysis in the New England Journal of Medicine, led by Patrick Turley, reached the same broad verdict for education. Selecting the best of ten embryos yielded an expected gain of about 0.53 years of schooling for people of European ancestry, and less for other ancestries.

The uncertainty dwarfs the effect. The Turley team calculated that the actual outcome for a selected embryo could range from more than four extra years of education to more than three fewer. The chosen embryo might end up doing worse, not better.

That interval is the point too often lost in the sales pitch. A prediction can be perfectly valid across thousands of children yet nearly useless for the one child a couple is planning for. That gap is where the whole controversy lives.

For disease, the same paper found small but real absolute reductions when selection was done well. For parents of European ancestry choosing among ten embryos, the expected lifetime risk of type 2 diabetes, 35.3 per cent in the United States, fell by about 5.5 percentage points. The risk of coronary artery disease, 6.7 per cent, fell by about 1.1 points.

The authors stressed that such reductions should always be shown in absolute terms. Their type 1 diabetes example shows why: a 35 per cent cut in relative risk sounds dramatic, but with a lifetime risk of only 0.34 per cent, it moves a child’s odds by 0.12 percentage points, which is a very different thing to buy.

How the embryos are chosen matters too. A 2021 analysis in the journal eLife, led by Todd Lencz, found that discarding only embryos with very high scores achieves almost nothing, while always choosing the lowest-scoring embryo can produce large relative reductions for a single disease — provided a couple has enough viable embryos, which many IVF patients do not.

The Nucleus Embryo Claims, Assessed

The company behind the product was founded in New York in 2021 by Kian Sadeghi, then in his early twenties, and has raised tens of millions of dollars from prominent technology investors. At launch it did not biopsy embryos itself, but partnered with an existing clinical genetic-testing firm.

Parents would upload embryo data after standard clinical screening, then receive rankings across hundreds of conditions and dozens of additional analyses.

The business has grown since launch. In November 2025 an advertising campaign in New York City urged people to “have your best baby”. The company now sells an IVF+ package, reported by CBS News at $30,000, and by mid-2026 had signed clinics in California, New York and Hawaii. In March 2026 it announced agreements with Indira IVF, India’s largest IVF chain, and with the Abu Ghosh Fertility Group in Jordan. In India, selecting embryos for non-medical traits is illegal, which confines what can be offered there to disease risk.

The company is also in court. In October 2025 Genomic Prediction, the testing laboratory Nucleus worked with at launch, sued it and Nathan Treff, Genomic Prediction’s former chief scientific officer, who had joined Nucleus, alleging misappropriation of trade secrets. Nucleus and Treff deny the claims, and the court declined Genomic Prediction’s request for emergency orders.

Nucleus Embryo marketed selection for resistance to heart disease and cancer, for intelligence and longevity, for body mass index, for eye and hair colour, and even for left-handedness and a predisposition to alcoholism. Each claim deserves its own scrutiny.

The Theranos Comparison and the Problem of Hype

The sharpest criticism of Nucleus Embryo is not that its technology is fake. It is that a genuine research tool has been repackaged into a consumer promise it cannot keep. That structural problem is what invites the Theranos comparison.

The parallel is imperfect. Nucleus uses real genomic technology, not fabricated results. But the shape of the risk rhymes: dramatic claims marketed to people willing to pay for control over something they care about, wrapped in science too complex for most buyers to check.

The stakes are unusually high because the customers are prospective parents and the product concerns their future children. A family may pay thousands, choose an embryo on a prediction that never materialises, then measure a real child against a number that was always shakier than it looked.

This is why the vocabulary matters. “Optimisation” implies a reliable dial the science does not provide. What the studies describe is a faint statistical nudge with an enormous margin of error — a very different thing from the confident language of a product page.

The Evidence Base That Barely Exists

The statistical weaknesses of polygenic embryo selection are set out above. There is a separate problem, and it is easier to check than any of the mathematics: nobody has run a trial of the outcomes that matter.

Searching ClinicalTrials.gov on 9 October 2026 turns up two registered studies of polygenic embryo selection. The first, registered in 2020 by Genomic Prediction, set out to measure how interested IVF patients were in receiving polygenic risk reports; its listing has not been updated since, and its status is unknown. The second, registered by Nucleus Genomics on 25 September 2026 and not yet recruiting, plans to compare about 2,000 patients whose embryos are chosen by polygenic score or by their appearance under the microscope. Its main outcome is whether the embryo implants and the pregnancy continues.

Neither study is randomised, both are run by companies that sell the test, and neither asks the fundamental question: do children selected this way actually differ, in health or in the traits on the report, from children who were not? That question has never been formally posed in a registered clinical study.

The comparison with neighbouring fields is instructive. The same registry holds 2,933 studies involving CAR-T cell therapy, 773 involving mRNA vaccines, 220 involving antisense oligonucleotides and 208 involving siRNA. Genetic medicine registers its trials. Even mitochondrial donation, a far rarer intervention, has 15. Polygenic embryo selection has two, neither designed to test its central claim, while being marketed directly to consumers.

There are honest reasons why such a trial is hard. The outcome you would want to measure — adult heart disease, adult educational attainment — arrives decades after the embryo is selected. A randomised design raises obvious ethical objections, since it would mean deliberately implanting embryos ranked lower by the very score under test. These are real obstacles.

But the difficulty of gathering evidence is an argument for caution in selling the service, not a licence to sell it without evidence. In most areas of medicine, a product whose central claim cannot yet be tested is not offered commercially in the meantime. Here the sequence has been inverted: the market arrived first, and the evidence has not followed.

Anyone being offered this should ask one question before anything else: what study, on how many children, followed for how long, supports the number on this report? At present there is no answer, because there is no such study.

What the Ethical Debate Is Really About

The deepest concerns are not about the precise capabilities of one product. They are about direction of travel: the gradual normalisation of choosing children by predicted traits, and the social consequences that could follow.

Demand is not hypothetical. In a survey of 6,823 American adults in January 2022, published in Science the following year, 58 per cent said they had no moral objection to polygenic embryo screening, and 38 per cent said they would be more likely than not to use it to raise their child’s chances of attending a top-100 college — if it were free, safe, and they were already going through IVF.

The history of eugenics is the essential context. In the early twentieth century, scientific and political establishments across several countries embraced selective reproduction to improve populations. In the United States, sterilisation laws led to the forced sterilisation of tens of thousands of people.

In Nazi Germany the same logic escalated catastrophically. The 1933 sterilisation law was followed, according to the United States Holocaust Memorial Museum, by the forced sterilisation of around 400,000 people, and then by an organised murder programme.

Modern genomic selection differs in crucial ways. It acts on embryos rather than living people, it is voluntary, and its science is more sophisticated. But the underlying idea — that some genetic profiles are more desirable — echoes that history enough to demand caution.

Access and equity sharpen the concern. Effective embryo selection would first reach only affluent families who can afford IVF plus extra screening. Over time, a society where advantaged families select for health and cognition while others cannot would let genetic inequality compound economic inequality across generations.

The ancestry gap makes this worse. Because the underlying scores work best for people of European descent, any advantage the technology confers would flow unevenly, mapping onto existing lines of privilege rather than cutting across them. A tool sold as personal empowerment could deepen structural inequality.

There is a quieter worry too. Parents who pay thousands for a genetic ranking may come to view a child through that lens, measuring a living person against a prediction. How that shapes the parent–child bond is hard to foresee and harder to undo.

Where the Law Draws the Line

Regulation varies dramatically by country. In the United Kingdom the Human Fertilisation and Embryology Authority has been unambiguous: polygenic trait screening of embryos, known as PGT-P, does not fall within the purposes permitted by law and is therefore unlawful.

The regulator adds that there is no evidence supporting the clinical use of such screening to improve either the chance of a baby or that child’s later health, and warns that it may even reduce the chance of having a baby at all. UK law also prohibits choosing an embryo’s sex for any reason other than avoiding a serious inherited illness.

The United States is far more permissive. Federal oversight of genetic tests is limited, resting mainly on laboratory quality standards and medical-device rules, with wide gaps for direct-to-consumer products, especially when a service presents itself as software rather than a diagnostic device.

That software framing is the crux. A product that merely presents information, rather than diagnosing or treating, can sidestep the stricter rules that would apply to a medical device — which is part of why a service unlawful in Britain can be sold openly in America.

Much of Europe sits closer to the UK. Germany, Spain, France, Italy and Norway restrict embryo selection to serious therapeutic purposes, which means a product like Nucleus Embryo cannot lawfully be offered in most of those countries.

Professional bodies have hardened their positions. The European Society of Human Genetics called polygenic embryo screening “an unproven, unethical practice” in 2022. The American College of Medical Genetics and Genomics said in 2024 that it should not be offered as a clinical service. And in December 2025 the American Society for Reproductive Medicine, whose members run many of the clinics involved, concluded that current evidence does not support the technique’s predictive accuracy, safety or clinical value, and that it should not be offered as a reproductive service at this time.

The safe, regulated frontier of the field is better represented by advances covered in our report on gene editing in 2026.

For now the honest summary is narrow. Preimplantation testing to prevent serious single-gene disease is genuine, valuable medicine. Ranking embryos by predicted intelligence or longevity is not, and no amount of confident marketing changes what the peer-reviewed evidence actually shows.

What Scientists Say

Writing in Scientific American, bioethicists Arthur Caplan of New York University and James Tabery of the University of Utah drew a pointed comparison between Nucleus Embryo and Theranos, the blood-testing firm that vastly overstated its technology.

They were careful to note there is no sign of the deliberate deception that defined Theranos. The parallel was structural — a real underlying technology stretched into claims that entice investors and wealthy customers but do not survive scrutiny.

“The real danger is that a bunch of wealthy parents-to-be who are too eager to control their children’s biological future will shell out $5,999 for a product that offers no such control.”
— Arthur Caplan & James Tabery, bioethicists. Scientific American, 2025.

Reproductive-medicine specialists have echoed the caution. The researchers behind the Cell and New England Journal of Medicine analyses above consistently find that predicting complex traits from embryo DNA delivers far less than the marketing implies.

Frequently Asked Questions

Sources

Primary peer-reviewed research:

1. Karavani, E., et al. (2019). Screening Human Embryos for Polygenic Traits Has Limited Utility. Cell, 179(6), 1424–1435. doi.org/10.1016/j.cell.2019.10.033
2. Turley, P., et al. (2021). Problems with Using Polygenic Scores to Select Embryos. New England Journal of Medicine, 385(1), 78–86. doi.org/10.1056/NEJMsr2105065
3. Okbay, A., et al. (2022). Polygenic prediction of educational attainment within and between families. Nature Genetics, 54(4), 437–449. doi.org/10.1038/s41588-022-01016-z
4. Martin, A. R., et al. (2019). Clinical use of current polygenic risk scores may exacerbate health disparities. Nature Genetics, 51(4), 584–591. doi.org/10.1038/s41588-019-0379-x
5. Herskind, A. M., et al. (1996). The heritability of human longevity: a study of 2872 Danish twin pairs. Human Genetics, 97(3), 319–323. doi.org/10.1007/BF02185763
6. Yan, J., et al. (2021). Live Birth with or without Preimplantation Genetic Testing for Aneuploidy. New England Journal of Medicine, 385, 2047–2058. doi.org/10.1056/NEJMoa2103613
7. Lencz, T., et al. (2021). Utility of polygenic embryo screening for disease depends on the selection strategy. eLife, 10, e64716. doi.org/10.7554/eLife.64716
8. Duncan, L., et al. (2019). Analysis of polygenic risk score usage and performance in diverse human populations. Nature Communications, 10, 3328. doi.org/10.1038/s41467-019-11112-0
9. Meyer, M. N., et al. (2023). Public views on polygenic screening of embryos. Science, 379, 541–543. doi.org/10.1126/science.ade1083
10. Shenhar, B., et al. (2026). Heritability of intrinsic human life span is about 50% when confounding factors are addressed. Science, 391, 504–510. doi.org/10.1126/science.adz1187

Institutional / science journalism:

11. American Society for Reproductive Medicine (2024). The use of preimplantation genetic testing for aneuploidy: a committee opinion. Fertility and Sterility. asrm.org
12. Human Fertilisation and Embryology Authority. PGT-P is not lawful in the UK. HFEA — embryo testing: what is and is not permitted in the UK
13. United States Holocaust Memorial Museum, Holocaust Encyclopedia. Eugenics. encyclopedia.ushmm.org
14. Caplan, A. & Tabery, J. (2025). The Myth of the Designer Baby. Scientific American. scientificamerican.com
15. ClinicalTrials.gov — registered studies of polygenic embryo selection (queried 9 October 2026: two company-sponsored observational studies)
16. Ethics and Practice Committees of the American Society for Reproductive Medicine (2026). Use of preimplantation genetic testing for polygenic disorders (PGT-P): an Ethics Committee opinion. Fertility and Sterility, 125, 24–30. asrm.org
17. Grebe, T. A., et al. (2024). Clinical utility of polygenic risk scores for embryo selection: a points to consider statement of the American College of Medical Genetics and Genomics. Genetics in Medicine, 26, 101052. doi.org/10.1016/j.gim.2023.101052
18. Forzano, F., et al. (2022). The use of polygenic risk scores in pre-implantation genetic testing: an unproven, unethical practice. European Journal of Human Genetics, 30, 493–495. doi.org/10.1038/s41431-021-01000-x
19. Human Fertilisation and Embryology Authority. PGT-P is not lawful in the UK, is not supported by evidence and may reduce the chances of having a baby overall. hfea.gov.uk
20. ClinicalTrials.gov. NCT07841769 — Prospective observational outcome comparison of morphology- versus polygenic-based selection of IVF-derived embryos (Nucleus Genomics, first posted 25 September 2026). clinicaltrials.gov
21. ClinicalTrials.gov. NCT04528498 — The Embryo Health Study: prospective longitudinal analysis of PGT-P (Genomic Prediction, 2020). clinicaltrials.gov