Here's the primary source https://iopscience.iop.org/article/10.3847/2041-8213/adc1c8 They possibly detected dimethyl sulfide, which is only known to be produced by living organisms.
I'm not convinced about the methods. It looks a lot like p-hacking to me: they have a highly specific hypothesis drawn from a large universe—that dozen or so molecules (§3.1) in their infrared spectrum model they're fitting experimental data against. I don't buy the way they created that hypothesis. The put a handful of highly specific biosignature gases into it, things that were proposed by exobiology theory papers.…
- Alien metabolites are a low-prior probability hypothesis. Dimethyl sulfide is a long-postulated biosignature with no natural source, so, it's low-prior
- The paper's model fits Webb data—a handful of photons—against no more than 20 candidate molecules, combined across all of their atmospheric models. Many of those gases are drawn from that low-prior "alien metabolite" class
- There's a much larger class of strongly infrared-absorbing gases, that can naturally occur in planetary atmospheres. Beyond those included in the 20 candidates. These (should!) have higher prior probability of occurring in Webb data than alien metabolites. (This class is so large and complicated, there's major spectral features in our own solar system's gas planets we haven't characterized yet)
- If you were to fit Webb data against that expanded class, those alternative hypotheses, you'd get a large number of 3-sigma detections by pure chance.
- The Webb data is too weak to distinguish between these. With only a few bits of information, you can distinguish between only a small set of alternative hypotheses
- This paper elevates the alien-metabolite hypothesis very highly, and that is why when it has a spurious statistical detection, it happens to be an alien metabolite detection. Because that hypothesis is overrepresented in their model
- The root problem is that since there's only a trickle of real data from this exoplanet, from Webb, it's unlikely one can infer anything super interesting from those few bits