Sign up for CNN’s Wonder Theory science newsletter. Explore the universe with news on fascinating discoveries, scientific advancements and more.
A tiny sign revealed in April seemed like it might change the universe as we know it.
Astronomers had detected just a hint, a glimmer of two molecules swirling in the atmosphere of a distant planet called K2-18b — molecules that on Earth are produced only by living things. It was a tantalizing prospect: the most promising evidence yet of an extraterrestrial biosignature, or traces of life linked to biological activity.
But only weeks later, new findings suggest the search must continue.
“It was exciting, but it immediately raised several red flags because that claim of a potential biosignature would be historic, but also the significance or the strength of the statistical evidence seemed to be too high for the data,” said Dr. Luis Welbanks, a postdoctoral research scholar at Arizona State University’s School of Earth and Space Exploration.
While the molecules identified on K2-18b by the April study — dimethyl sulfide, or DMS, and dimethyl disulfide, or DMDS — are associated largely with microbial organisms on our planet, scientists point out that the compounds can also form without the presence of life. Now, three teams of astronomers not involved with the research, including Welbanks, have assessed the models and data used in the original biosignature discovery and got very different results, which they have submitted for peer review.
Meanwhile, the lead author of the April study, Nikku Madhusudhan, and his colleagues have conducted additional research that they say reinforces their previous finding about the planet. And it’s likely that additional observations and research from multiple groups of scientists are on the horizon.
The succession of research papers revolving around K2-18b offers a glimpse of the scientific process unfolding in real time. It’s a window into the complexities and nuances of how researchers search for evidence of life beyond Earth — and shows why the burden of proof is so high and difficult to reach.
Noisy data
Located 124 light-years from Earth, K2-18b is generally considered a worthy target to scour for signs of life. It is thought to be a Hycean world, a planet entirely covered in liquid water with a hydrogen-rich atmosphere, according to previous research led by Madhusudhan, a professor of astrophysics and exoplanetary science at the University of Cambridge’s Institute of Astronomy. And as such, K2-18b has rapidly attracted attention as a potentially habitable place beyond our solar system.
Convinced of K2-18b’s promise, Madhusudhan and his Cambridge colleagues used observations of the planet by the largest space telescope in operation, the James Webb Space Telescope, to study the planet further. But two scientists at the University of Chicago — Dr. Rafael Luque, a postdoctoral scholar in the university’s department of astronomy and astrophysics, and Michael Zhang, a 51 Pegasi b / Burbidge postdoctoral fellow — spotted some problems with what they found.
After reviewing Madhusudhan and his team’s April paper, which followed up on their 2023 research, Luque and Zhang noticed that the Webb data looked “noisy,” Luque said.
Noise, caused by imperfections in the telescope and the rate at which different particles of light reach the telescope, is just one challenge astronomers face when they study distant exoplanets. Noise can distort observations and introduce uncertainties into the data, Zhang said.
Trying to detect specific gases in distant exoplanet atmospheres introduces even more uncertainty. The most noticeable features from a gas like dimethyl sulfide stem from a bond of hydrogen and carbon molecules — a connection that can stretch and bend and absorb light at different wavelengths, making it hard to definitively detect one kind of molecule, Zhang said.
“The problem is basically every organic molecule has a carbon-hydrogen bond,” Zhang said. “There’s hundreds of millions of those molecules, and so these features are not unique. If you have perfect data, you can probably distinguish between different molecules. But if you don’t have perfect data, a lot of molecules, especially organic molecules, look very similar, especially in the near-infrared.”
Delving further into the paper, Luque and Zhang also noticed that the perceived temperature of the planet appeared to increase sharply from a range of about 250 Kelvin to 300 Kelvin (-9.67 F to 80.33 F or -23.15 C to 26.85 C) in research Madhusudhan published in 2023 to 422 Kelvin (299.93 F or 148.85 C) in the April study.
Such harsh temperatures could change the way astronomers think about the planet’s potential habitability, Zhang said, especially because cooler temperatures persist in the top of the atmosphere — the area that Webb can detect — and the surface or ocean below would likely have even higher temperatures.
“This is just an inference only from the atmosphere, but it would certainly affect how we think about the planet in general,” Luque said.
Part of the issue, he said, is that the April analysis didn’t include data collected from all three Webb instruments Madhusudhan’s team used over the past few years. So Luque, Zhang and their colleagues conducted a study combining all the available data to see whether they could achieve the same results, or even find a higher amount of dimethyl sulfide. They found “insufficient evidence” of both molecules in the planet’s atmosphere.
Instead, Luque and Zhang’s team spotted other molecules, like ethane, that could fit the same profile. But ethane does not signify life.
Disappearing evidence
Arizona State’s Welbanks and his colleagues, including Dr. Matt Nixon, a postdoctoral researcher in the department of astronomy at the University of Maryland College Park, also found what they consider a fundamental problem with the April paper on K2-18b.
The concern, Welbanks said, was with how Madhusudhan and his team created models to show which molecules might be in the planet’s atmosphere.
“Each (molecule) is tested one at a time against the same minimal baseline, meaning every single model has an artificial advantage: It is the only explanation permitted,” Welbanks said.
When Welbanks and his team conducted their own analysis, they expanded the model from Madhusudhan’s study.
“(Madhusudhan and his colleagues) didn’t allow for any other chemical species that could potentially be producing these small signals or observations,” Nixon said. “So the main thing we wanted to do was assess whether other chemical species could provide an adequate fit to the data.”
When the model was expanded, the evidence for dimethyl sulfide or dimethyl disulfide “just disappears,” Welbanks said.
Burden of proof
Madhusudhan believes the studies that have come out after his April paper are “very encouraging” and “enabling a healthy discussion on the interpretation of our data on K2-18b.”
He reviewed Luque and Zhang’s work and agreed that their findings don’t show a “strong detection for DMS or DMDS.” When Madhusudhan’s team published the paper in April, he said the observations reached the three-sigma level of significance, or a 0.3% probability that the detections occurred by chance.
For a scientific discovery that is highly unlikely to have occurred by chance, the observations must meet a five-sigma threshold, or below a 0.00006% probability that the observations occurred by chance. Meeting such a threshold will require many steps, Welbanks said, including repeated detections of the same molecule using multiple telescopes and ruling out potential nonbiological sources.
While such evidence could be found in our lifetime, it is less likely to be a eureka moment and more a slow build requiring a consensus among astronomers, physicists, biologists and chemists.
“We have never reached that level of evidence in any of our studies,” Madhusudhan wrote in an email. “We have only found evidence at or below 3-sigma in our two previous studies (Madhusudhan et al. 2023 and 2025). We refer to this as moderate evidence or hints but not a strong detection. I agree with (Luque and Zhang’s) claim which is consistent with our study and we have discussed the need for stronger evidence extensively in our study and communications.”
In response to the research conducted by Welbanks’ team, Madhusudhan and his Cambridge colleagues have authored another manuscript expanding the search on K2-18b to include 650 types of molecules. They have submitted the new analysis for peer review.
“This is the largest search for chemical signatures in an exoplanet to date, using all the available data for K2-18b and searching through 650 molecules,” Madhusudhan said. “We find that DMS continues to be a promising candidate molecule in this planet, though more observations are required for a firm detection as we have noted in our previous studies.”
Welbanks and Nixon were pleased that Madhusudhan and his colleagues addressed the concerns raised but feel that the new paper effectively walks back central claims made in the original April study, Welbanks said.
“The new paper tacitly concedes that the DMS/DMDS detection was not robust, yet still relies on the same flawed statistical framework and a selective reading of its own results,” Welbanks said in an email. “While the tone is more cautious (sometimes), the methodology continues to obscure the true level of uncertainty. The statistical significance claimed in earlier work was the product of arbitrary modeling decisions that are not acknowledged.”
Luque said the Cambridge team’s new paper is a step in the right direction because it explores other possible chemical biosignatures.
“But I think it fell short in the scope,” Luque said. “I think it restricted itself too much into being a rebuttal to the (Welbanks) paper.”
Separately, however, the astronomers studying K2-18b agree that pushing forward on researching the exoplanet contributes to the scientific process.
“I think it’s just a good, healthy scientific discourse to talk about what is going on with this planet,” Welbanks said. “Regardless of what any single author group says right now, we don’t have a silver bullet. But that is exactly why this is exciting, because we know that we’re the closest we have ever been (to finding a biosignature), and I think we may get it within our lifetime, but right now, we’re not there. That is not a failure. We’re testing bold ideas.”
For more CNN news and newsletters create an account at CNN.com