Science News, March 25

Climate science’s new red alert: the world has just logged its hottest 11-year stretch on record

The most consequential science headline of the week is not a single experiment but a planetary status update. The World Meteorological Organization’s State of the Global Climate 2025, released on 23 March 2026, says 2015–2025 were the hottest 11 years ever recorded and that 2025 was the second or third warmest year on record, at about 1.43°C above the 1850–1900 average. The report’s most striking addition is a new headline metric: Earth’s energy imbalance, which hit a record in 2025. In plain terms, more heat is entering the Earth system than leaving it, and over 91% of that excess heat is going into the oceans, while sea ice and glaciers keep shrinking. This matters because it turns climate change from an abstract temperature trend into a systems-level warning about food, health, water, sea level, and extreme weather. Of all the stories this week, this is the one with the deepest long-term consequences for every country, including Romania. (World Meteorological Organization)

CERN moved antimatter outside the lab for the first time

One of the week’s most iconic physics milestones came from CERN: scientists transported trapped antimatter—specifically 92 antiprotons—in a specially designed portable trap loaded onto a truck. That sounds almost absurdly small, but it is a major technical breakthrough because antimatter annihilates on contact with ordinary matter and has to be held in extreme vacuum and electromagnetic confinement. CERN says the achievement opens the door to moving antimatter away from the noisy environment of particle accelerators and into other laboratories where it can be studied more precisely. Why that matters: antimatter experiments are among the cleanest ways to test fundamental symmetries in physics and probe one of the biggest mysteries in cosmology—why the Universe today is dominated by matter rather than equal parts matter and antimatter. This story is important not because of the quantity moved, but because it transforms antimatter from something that must stay “where it is made” into something researchers may eventually use more flexibly across Europe. (home.web.cern.ch)

Regenerative medicine took a big step with a lab-grown oesophagus that worked in pigs

A medical advance from 20 March could become one of the most important translational biology stories of the month. Researchers reported that bioengineered oesophageal tissue restored swallowing in minipigs after implantation, raising the prospect of future treatment for children born with oesophageal defects and for adults whose tissue has been damaged by cancer. The underlying Nature Biotechnology paper describes an autologous engineered esophagus built on a decellularized porcine scaffold and seeded with the animals’ own precursor cells to repair 2.5-centimetre circumferential defects. The significance is twofold. First, the oesophagus is mechanically and biologically demanding tissue: it has to heal, remain open, and move food reliably. Second, this is not just a proof-of-concept organoid in a dish; it is a large-animal result in a structure relevant to surgery. Human use is still ahead, and there are many safety and durability questions left. But in regenerative medicine, moving from “cells grow in the lab” to “functional tissue works in a living large animal” is exactly the kind of threshold scientists watch for. (Nature)

AI-generated medical images are now realistic enough to fool radiologists

Another major story this week is less about discovery and more about a looming scientific and clinical risk. A study highlighted by Nature found that deepfake chest X-rays were hard to distinguish from real ones not just for general AI tools but even for specialists: 17 radiologists from 12 research centres struggled, with fewer than half reliably spotting the synthetic images. The immediate implication is obvious: medical imaging can no longer be treated as inherently trustworthy visual evidence. That matters for clinical workflows, model training, insurance, fraud, and even court cases where imaging is introduced as proof of injury. The longer-term implication is that hospitals and imaging archives may need the equivalent of cybersecurity for pictures: provenance, watermarking, audit trails, and better forensic detection tools. This story deserves a place near the top because it shows how fast generative AI is moving from “funny internet problem” to “health-system vulnerability.” It is a science story, but also a warning that the boundary between digital fabrication and biological evidence is collapsing faster than many regulators expected. (Nature)

A 20-year cloning experiment found a hard biological limit: clones of clones eventually fail

A team in Japan answered a question that has hovered over mammalian cloning since Dolly: could cloning be repeated forever if each clone became the donor for the next one? The answer now appears to be no. In a study released on 24 March, researchers reported that after 20 years of serial cloning from a single donor mouse, the line collapsed at the 58th generation. The birth rate began falling after the 27th generation, and by the end, all 58th-generation clones died shortly after birth. Reuters reports the team produced 1,206 cloned mice from the original donor; genomic analysis indicated harmful mutations accumulated over time, undermining the old assumption that repeated cloning could continue indefinitely without severe degradation. This is important beyond mouse genetics. It sharpens the limits of cloning as a tool for conservation, agriculture, and biomedical replication. It also reinforces an old evolutionary lesson: sexual reproduction is biologically messy, but it helps prevent the silent accumulation of damage that a pure copy-of-a-copy system seems unable to escape. (Nature)

Perseverance found some of the oldest evidence yet of flowing water on Mars

NASA’s Perseverance rover delivered one of the week’s most compelling planetary-science results. Using its RIMFAX ground-penetrating radar, the rover detected buried remains of an ancient river delta beneath Jezero Crater, with sediments and eroded structures dating to roughly 3.7 to 4.2 billion years ago. That makes the feature older than the already-known nearby Western Delta. Why scientists care is simple: deltas are ideal places to trap sediments, preserve chemical clues, and potentially hold biosignatures from ancient microbial life. The result does not prove life ever existed on Mars, but it strengthens the case that Jezero was once a water-rich environment capable of preservation, and that Mars’s wet past may have begun even earlier than some surface features suggested. It also shows the value of looking below the surface rather than only reading exposed rocks. In Mars exploration, each new buried structure matters because habitability is a geological story first: before you can ask whether life existed, you have to understand where water pooled, where sediments accumulated, and where delicate evidence could have survived. (Reuters)

CERN also announced a new heavy proton-like particle, deepening the picture of the strong force

Transporting antimatter was not CERN’s only big week. The LHCb collaboration announced the discovery of a new particle, Ξcc⁺, described as a proton-like particle made of two charm quarks and one down quark. Because charm quarks are heavy, the particle’s mass is about four times that of a normal proton. This is not the kind of headline that changes everyday life tomorrow, but in fundamental physics it matters a great deal. Particles like this are laboratories for understanding quantum chromodynamics, the theory of the strong force that binds quarks into protons, neutrons, and other hadrons. Systems containing two heavy quarks are especially valuable because they test how quarks behave in combinations that are rare in ordinary matter. CERN also noted this is the first new particle identified with the upgraded LHCb detector from Run 3-era data, which makes it a technological milestone as well as a scientific one. It is a reminder that even after the Higgs discovery, the LHC is still producing genuinely new building blocks for precision particle physics. (home.web.cern.ch)

Synthetic biology crossed another line with “zombie cells” revived by a donor genome

This is the boldest story on the list, but also the one that needs the biggest caveat: it is based on a bioRxiv preprint, not yet peer-reviewed. Researchers reported that they could take bacterial cells whose own DNA had been effectively destroyed—described by Nature as “dead” cells—and revive them by inserting the working genome of another species. If the result holds up, it is a powerful advance in whole-genome transplantation and could expand the toolkit for synthetic biology well beyond genome editing. The deeper conceptual point is what makes the story so striking: it reframes “life” less as the state of an intact cell and more as the presence of a functioning, replicating genome inside suitable cellular machinery. That has implications for minimal cells, bioengineering, and the design of organisms with customized genetic programs. It does not mean scientists are resurrecting multicellular organisms or creating life from scratch in a science-fiction sense. But it does suggest that the boundary between a non-functioning biological shell and a living cell is more programmable than many people assumed. (Nature)

Romania: UTCN launched what it calls the country’s first academic AI career-and-faculty navigator

From Romania, one of the freshest and most relevant science-and-technology stories came from Cluj on 23 March. Researchers at the North University Center of Baia Mare, part of the Technical University of Cluj-Napoca (UTCN), launched Navigatorul de Carieră & Facultate, a digital application designed to help high-school graduates choose studies and career paths. According to UTCN, it is a national premiere in IT and the first intelligent agent developed in a Romanian academic framework, with technical support from Holisun and availability through the ChatGPT interface. The recommendations focus mainly on Romanian universities, but also include European and global options. This is not a “breakthrough science” headline on the scale of CERN or Mars, but it is still important locally because it shows Romanian academia moving from talking about AI to building applied public-facing tools. If it works well, it could become a model for how Romanian universities use conversational AI not just for teaching assistance, but for guidance, access, and better alignment between education choices and labor-market realities. (AGERPRES)

Romania: non-invasive geophysics uncovered a forgotten 19th-century German cemetery in Târgu Neamț

Romania’s second standout science story in the last few days is an archaeology one. On 21 March, AGERPRES reported that researchers from Friedrich-Alexander University Erlangen-Nürnberg, working with archaeologists from Complexul Muzeal Național Neamț, identified a 19th-century German cemetery near Târgu Neamț. What makes the story notable is not only the discovery itself, but the method: the team used modern non-invasive geophysical scanning to investigate structures below ground after old 19th-century maps hinted that a forgotten cemetery once existed there. The location had no obvious surface markers and is now basically pastureland, which means the find is a strong example of how archaeology is increasingly driven by remote sensing, subsurface imaging, and careful historical cartography rather than only excavation. For Romania, that matters because it shows heritage science operating at a high methodological level and recovering local history that had nearly vanished from public memory. It is a small-scale story compared with planetary or biomedical news, but a scientifically serious one with clear cultural value. (AGERPRES)