A study published in Nature on 2 September reports that semaglutide, started late in life, slowed several measures of ageing in mice and extended how long they lived. The comparison arm was calorie restriction — the one intervention that has reliably stretched rodent lifespan for the better part of a century — and semaglutide matched it on most measures and beat it on a few. Nothing here changes what any person should do this week. It does explain why “GLP-1 for longevity” is about to become a much louder sales pitch.
What the study did
Researchers at the University of California, Berkeley, with collaborators at the University of Copenhagen and the National Institute on Aging, treated 20-month-old female C57BL/6 mice with semaglutide for three months. Twenty months is late middle age for that strain, so this was not a lifelong intervention: treatment began when the animals were already old. Over those three months the treated mice showed improved physiological function, reduced hallmarks of ageing, and shifts in nutrient sensors and in the conserved genes that regulate ageing across species. Continued treatment extended lifespan.
The more informative arm was a longitudinal head-to-head against matched calorie restriction. Semaglutide preserved baseline function and reproduced many of the functional benefits of calorie restriction by slowing age-related decline. On three measures — exploratory drive, spatial memory and glucose control — it did better than calorie restriction, producing improvements above the animals’ own baseline rather than merely a gentler slide downwards. The authors’ conclusion is that GLP-1 receptor activation works as a calorie-restriction mimetic, which would go some way to explaining why these drugs keep producing benefits in organs with no obvious connection to appetite.
One number the published abstract does not give is the size of the lifespan extension. There is no percentage and no median survival figure. That number matters a great deal, and it is not yet in front of us.
Why this reaches marketing before it reaches the clinic
GLP-1 drugs have accumulated an odd list of wins outside weight and glucose: kidney outcomes in type 2 diabetes, resolution of metabolic dysfunction-associated steatohepatitis, reductions in cardiovascular events. Each was found empirically, one trial at a time, with no unifying mechanism underneath. A calorie-restriction mimetic is a mechanism, and if it holds up it reframes the entire class.
It is also the most commercially attractive claim available in medicine, and the translation record is poor. Rapamycin and metformin both extended rodent lifespan; neither has a completed human lifespan trial, and metformin’s TAME study has spent years short of funding. Mouse longevity results have a long history of not surviving contact with people.
The practical consequence for readers here is narrow and defensive. Nothing in this paper supports taking semaglutide for ageing, supports any particular dose, or validates a single product sold on anti-ageing claims — and the market in peptide products sold on longevity marketing will be citing it within days. A Nature paper about mice is not evidence for a vial bought online. Our account of what GLP-1 drugs are established to do in people is unchanged.
Four limits are worth carrying forward. Only female mice were studied, and sex differences in lifespan experiments are both common and large. The magnitude of the lifespan gain is not stated in the abstract. C57BL/6 laboratory mice on standard chow in a clean facility are a poor stand-in for a human population with varied diets and accumulated illness. And the Regents of the University of California have filed a patent application on GLP-1 receptor agonists for healthy ageing, which the authors disclose — not disqualifying, but it means the finding arrives attached to a commercial interest. The paper is published online ahead of print.
Sources: Feng Y, Barthez M, Wang Y, et al. Late-life semaglutide treatment slows ageing and extends lifespan in female mice. Nature, published online 2 September 2026, doi:10.1038/s41586-026-10940-7 · PubMed record, PMID 42686906
