Choose the "Best" Embryo, Then Rewrite It: The New Baby Business Is Open

Two frontier technologies are meeting in the fertility clinic: one to rank your future children, one to edit them. Here is what is real, what is oversold, and the fight over whether we should do it at all.

Picture a couple in a fertility clinic, staring at a screen. Five embryos sit in front of them, each a tiny cluster of frozen cells. A few years ago the doctor could say almost nothing about them. Now a report ranks all five on scores drawn from across their DNA: Embryo 3 carries a lower predicted risk of heart disease, and scores a touch higher on "cognitive potential."

Which would you pick? And once that question feels normal, how long before someone offers to fix the ones you did not choose?

This is not science fiction. Two fast-moving technologies are converging on the IVF clinic, and together they push toward something once unthinkable: the power to select a future child, and maybe to rewrite one. Choose, then edit. That is the whole story, and it is arriving faster than the debate around it.

The first is embryo screening. Startups sequence an embryo's DNA and rank a batch by "polygenic scores," bets stitched together from thousands of tiny genetic markers across the genome. The industry calls it PGT-P, preimplantation genetic testing for polygenic conditions. In January 2026, MIT Technology Review named embryo scoring one of its ten Breakthrough Technologies of the year, a sign it had left the fringe.

Reading an embryo's genome is hard, because there are so few cells. Herasight sequences the parents deeply and reconstructs each embryo from their data; Orchid sequences each embryo directly. The money is not shy, with backers including Peter Thiel and Alexis Ohanian. The label promises a lower chance of passing on cancer, diabetes or heart disease. The subtext is louder: a future where you shop for the traits of your child.

Predicting a complex trait from DNA is a matter of probability, and for many traits it is weak. Height, health and intelligence are shaped by thousands of genes, tangled with environment, luck and upbringing. A polygenic score captures only a slice, and for the traits people most want to shop for, that slice is thin. For a trait like ADHD, the best published score explains only about one percent of why people differ, by Herasight's own analysis.

There is even a public accuracy fight inside the industry: Herasight claims a rival, Nucleus Genomics, overstates what its scores can support, and Nucleus disputes it. Critics of the "rank your baby by IQ" pitch have a blunter word: pseudoscience. Accuracy also varies by ancestry, since most of the data comes from people of European descent. None of this makes the scores useless; for serious single-gene conditions, testing embryos is established medicine. The problem is the leap from "lower risk of one disease" to "smarter, taller, better," sold with a confidence the numbers do not earn.

If scoring is choosing from the hand you were dealt, editing changes the cards. A cluster of startups wants to edit embryos' DNA directly, among them Preventive, backed by Sam Altman and Brian Armstrong, and Bootstrap Bio. Their tool is base editing, a precise cousin of CRISPR. Where ordinary CRISPR cuts the DNA strand and lets the cell repair it, base editing changes a single letter of code without cutting at all, which makes it more precise and, in theory, safer.

One catch towers over the rest: editing human embryos to start a pregnancy is banned in the United States and much of the world, so some of this work is pushed toward looser jurisdictions, the United Arab Emirates named repeatedly. And the engineering says slow down anyway. You can make only a handful of edits in a day-old embryo, not thousands, and the traits people fantasise about, intelligence and lifespan, are the most polygenic of all. A few edits cannot move them. Worse, base editing can still land in the wrong place, and because the change is made in an embryo, any mistake is inherited by every generation that follows. Not private. Permanent.

Strip away the specifics and one question remains, the most socially explosive in biology: should anyone hold the power to select, and maybe rewrite, a future child? The appeal is real: parents who carry a devastating disease would give anything to spare their child, and as a way to prevent suffering, this has a genuine moral pull. But the objections are serious too. If the "best" embryos can be bought, wealth buys a head start written into biology itself. And there is the loaded, historically dangerous idea that some lives are more worth choosing than others.

Frontier Signal will not hand you a verdict. Here is the honest shape of it. Embryo scoring is real, commercial and already sold, but its trait predictions, especially for IQ, are scientifically shaky, disputed within the industry, and skewed by whose DNA trained the models. Embryo editing is earlier still: pre-clinical, banned across much of the world, limited to a few edits, and pushed offshore. If you ever sit in that clinic holding one of these reports, the most valuable thing you can bring is not a score. It is one sharp question: how confident is this number, really? A ranking is a probability, not a promise. The baby business is open. Whether we should shop there at all, we have barely begun to answer.

Embryo screening reads an IVF embryo's DNA and ranks a batch by "polygenic scores," which add up the tiny effects of many genetic markers at once. In the trade it is called PGT-P. It estimates the odds of conditions such as heart disease, and, more controversially, traits. A score is a probability across a population, not a prediction about any one child.

Companies do sell scores that rank embryos on intelligence, but the science is thin and disputed. IQ is shaped by thousands of genes tangled with environment, so a score captures only a small slice, often a tiny one. Most models were trained mainly on people of European descent, making them less reliable for everyone else. Reading an embryo for a serious single-gene disease is established medicine; ranking it by predicted IQ is not.

Editing human embryos to start a pregnancy is banned in the United States and much of the world, and it remains pre-clinical everywhere. A few startups are exploring it and point to more permissive places such as the United Arab Emirates, but no clinic offers it. It can currently make only a few edits, and any change would be inherited by future generations.

Related from Frontier Signal: the same gene-editing toolkit is now approved to cure sickle cell in children, editing to cure rather than to enhance.

Frontier Signal explains frontier technology in plain English. This is general information, not medical advice.