In a high-level report precipitated by the birth of CRISPR babies in China in 2018, scientists say the technology’s next medical use should be narrowly restricted to prospective parents who can’t otherwise have a healthy child, such as Black couples who both have sickle-cell disease.

The 200-plus-page report, from the US National Academies and the UK Royal Society, says “heritable genome editing,” or the use of powerful DNA-editing tools like CRISPR to rewrite the genes of human embryos, is not yet safe enough to use in IVF clinics, but anticipates it could become so in the future.

If it does get used again, the technology’s initial application should be to help couples with no chance of having a healthy child otherwise, a rare situation but one in which the benefits could most clearly outweigh the risks.

In a Zoom call with journalists, Richard P. Lifton, the panel’s co-chairman and the president of Rockefeller University, in Manhattan, described the report as diagramming a “staged rollout of the technology, to make sure [it] is safe.” The panelists said futuristic applications like enhancing intelligence, or making kids with increased disease resistance, remain too uncertain to attempt.

Ever since CRISPR was first developed, scientists hoped to self-regulate the technology’s use and avoid legal bans. That suffered a setback in November 2018 after a little-known Chinese scientist, He Jiankui, said he had gone ahead and gene-edited twin embryos to make the baby girls resistant to HIV.

Widely denounced and later sentenced to a three-year prison term for violating medical regulations, He embarrassed scientific leaders. He’d told many of his plans and even cited an earlier report from the National Academies as his ethical green light. That 2017 report had called use of the technology premature but endorsed it as morally permissible.

The China debacle directly “precipitated” the current project, says Kay Davies, a geneticist at Oxford University and co-chair of the new panel. Unlike previous reports, this one does not dwell on ethical debates, but instead tackles the practical question of a “responsible translational pathway”—that is, what the technology’s appropriate use in medicine might be.

“Everyone needs to pay attention to the initial uses and how they should be limited. That is what this is about,” says Jeffrey Kahn, director of the Berman Institute of Bioethics at Johns Hopkins University, who sat on the panel.

Kahn, Davies, Lifton, and the other panelists said that if the technology matures and countries allow its use, the first applications should be in those rare situations in which two partners suffer from the same serious inherited disease. In such cases, neither parent would have a normal gene to pass down.

One example would be two people with cystic fibrosis who want to have kids. Such situations are not common, and the panel tried to estimate how often they could arise. For instance, in the US, about one in 13 African-Americans carries the genetic trait for sickle-cell disease. Those who inherit two copies of the mutation (about one in 350) will develop the condition, which affects the red blood cells and can cause painful, life-shortening complications. The report estimates there could be 80 couples in the US where both partners have sickle-cell disease.

Those couples would not be able to have healthy children of their own without genetic editing. In Asia, meanwhile, the blood disorder beta thalassemia is common enough to create similar situations, according to the report, and other such problems can arise in cultures where people frequently marry close cousins.

Jeanne O’Brien, a fertility specialist at Shady Grove Fertility in Maryland, says the panel’s focus on reproductive gene editing for Black people could raise questions about fairness. People with sickle-cell disease in the US “have been left behind because they are often Black and poor,” she says, but now they are sought after to participate in gene-therapy experiments. “It’s ironic this group is currently the one in demand for biotech innovation,” she says.

The scientists who wrote the report still don’t believe that genome editing is ready to use in reproductive clinics. That’s because it’s prone to introducing unexpected mutations that are hard to spot, and it can generate embryos with a mixture of edited and unedited cells. Those problems, and more, were apparent in an unpublished manuscript describing China’s CRISPR babies, which MIT Technology Review published excerpts of.

According to Benjamin Hurlbut, a sociologist at the Arizona State University, the new project was undertaken “to salvage the damaged reputation of the genome editing field by showing how to do responsibly what [He Jiankui] did irresponsibly.”

Using CRISPR safely in reproductive clinics will require technical improvements. One is a means of distinguishing normal from diseased embryos, even when they consist of a single cell, so that the former aren’t “needlessly” edited and brought to term. Also needed are more precise editing methods able to reproduce exact DNA sequences found in healthy people.

According to Eric Lander, head of the Broad Institute and a member of the panel, the group’s conclusion that genome-editing technology needs several more years of research amounts to an extended moratorium on further CRISPR babies. “The notion of clear thresholds and careful consideration about whether to cross them is an important aspect of the report,” he says.

In addition to pinpointing medically legitimate reasons to make CRISPR babies, the panelists also called for an international body to oversee the work. That could be something along the lines of the International Atomic Energy Agency, which inspects nuclear sites and polices treaties. Also needed is a way for whistleblowers to denounce unethical experiments, the panel said.

“Our group was very concerned about the potential for rogue scientists to embark on their own,” says Lifton. “This field is obviously moving very rapidly, with a high pace of technological advance. We can’t stay abreast of it without a mechanism.”

Whether would-be baby makers or governments will heed the recommendations is not clear. But many countries already ban the creation of genetically modified humans, whatever their purpose: the procedure is prohibited in the US and at least 66 other countries, according to the results of an unpublished survey by Françoise Baylis, a bioethicist at Dalhousie University, in Canada. Of more than 100 nations whose laws and regulations she reviewed, only a handful would permit the technology’s use.