Study Suggests AI-Generated Drug Could Slow Aging

A study published in Nature Biotechnology reports that Insilico Medicine's AI-generated drug rentosertib reduced biological age markers in 42 patients over a 12-week trial, as measured by six aging clocks, while placebo or untreated groups saw no improvement. The findings mark an early step toward using AI to develop anti-aging treatments, though the drug requires further trials before regulatory approval.

An AI-generated drug showed early signs of potentially slowing aging, according to a new study https://www.nature.com/articles/s41587-026-03286-y Abs1 out today in Nature Biotechnology, possibly taking the first small step towards using AI to cure all the diseases tech companies companies keep talking about. The new study centers on rentosertib, a drug from Insilico Medicine https://insilico.com , an AI drug discovery and development company. From the onset, Insilico said it created https://www.nytimes.com/2026/09/07/science/ai-generated-drug-longevity.html rentosertib with the aim of both treating patients suffering chronic lung disease and extending their lifespan. In the study https://www.nature.com/articles/s41587-026-03286-y Abs1 , the company showed that its drug managed to successfully reduce some of the biological markers of age, as measured by aging clocks, in a small set of patients. Aging clocks have moved from a sort of shorthand to a reference of where a person is in their life “their biological clock is ticking” to a real metric that aims to measure a person’s morbidity and mortality. A number of these clocks have been developed with AI to take many measurements—biomarkers in blood, physiological metrics, psychological assessments, etc.—and spit out a biological “age” meant to reflect a person’s overall health. In the 42 participants of the study, Insilico found that its drug managed to reduce their biological ages over the course of the 12-week trial, while those who received a placebo or no treatment saw their age increase or stay the same. The company used six different so-called aging clocks and noticed enough of a shift on all six to conclude that “the drug may modulate complementary dimensions of biological aging.” The idea of reducing aging certainly sounds good and has become the obsession of some tech bro-types https://gizmodo.com/bryan-johnsons-new-list-of-adorably-obvious-longevity-tips-shows-hes-come-a-long-way-2000759066 , but the actual utility of aging clocks is up for debate. There are several variations of these clocks, all of which take different approaches. The earliest versions, developed over a decade ago, track methylation https://tallyhealth.com/blogs/learn/dna-methylation-and-aging-how-epigenetic-clocks-measure-your-biological-age , which observes chemical changes in DNA. More recent models focus on proteomics, or measuring blood proteins, which a 2024 study https://www.nature.com/articles/s41591-024-03164-7 found could be used to determine if a person had a higher risk of age-related diseases. And while some researchers believe that these catch-all measurements could be useful https://www.nature.com/articles/s41514-025-00312-2 , there’s still a long way to go before they become truly meaningful markers. Scientists have tried to kick-start the field and improve accuracy https://www.science.org/content/article/scientific-showdown-seeks-biological-clock-best-tracks-aging so that such tools can be used for research purposes, but there still isn’t a measurement that serves as a universally accepted “age” indicator—and some doctors have warned that leaning on these metrics could lead to pursuing the wrong interventions https://www.aging-us.com/news-room/why-epigenetic-clocks-may-fail-to-measure-anti-aging-effects , seeking ways to slow aging indicators in a way that actually harms a body’s repair response. All that said, a pill that keeps a person’s body from succumbing to the effects of aging would be a breakthrough. Insilico’s pill is still a long way from being that—it’ll have lot of trials to go before getting any sort of regulatory approval.