American astronomers, supercharged by machine-learning tools, have uncovered the largest single trove of exoplanet candidates ever reported—more than 10,000 new worlds detected in imagery from NASA’s Transiting Exoplanet Survey Satellite (TESS). The Princeton-led team trained a Random Forest classifier to sift 80 million light curves, compressing a manual vetting task that would have taken over three years into a few weeks. The haul, dominated by hot Jupiters but also containing 11 potential super-Earths, demonstrates how AI is reshaping the pace of discovery. Separately, a decades-long debate around Barnard’s Star, the second-nearest star system to the Sun, was settled when spectrometers on Gemini North and the Very Large Telescope confirmed four rocky planets, each with a mass between 19 and 34 per cent of Earth’s, packed into tight orbits far hotter than the habitable zone.

While planet hunters filled out the exoplanetary census, other teams closed a different kind of inventory. An international collaboration using the CHIME radio telescope and the Dark Energy Spectroscopic Instrument located the “missing” ordinary matter that cosmological models predict but that had eluded detection inside galaxies. By measuring how fast radio bursts (FRBs) smear as they traverse intergalactic space, the researchers mapped diffuse gas extending up to 4 million light-years from galactic centres—far beyond the reach of previous simulations—and attributed its dispersal to black-hole and stellar feedback more violent than expected. Closer to home, a British-led study employing the Hubble Space Telescope’s ultraviolet spectrograph directly detected four white dwarfs orbiting red dwarf companions within 65 light-years of Earth. The objects had remained hidden because their cool, faint glow was drowned out in visible light; the observations revise the local white-dwarf space density upward and suggest up to ten similar binaries may still lurk undetected.

Chinese researchers contributed two findings with practical implications for exploration. Data from the Chang’e-6 mission revealed that the solar wind decelerates over the Moon’s near side, likely deflected by crustal magnetic anomalies—a detail that will inform radiation shielding for future crewed stations. In the deep ocean, scientists at the Chinese Academy of Sciences sequenced the genome of the isopod Bathynomus jamesi, a crustacean that survives more than five years without food at abyssal depths. A horizontally transferred bacterial gene, ND1, slows mitochondrial activity in cold conditions, boosting survival in zebrafish by 37 per cent in laboratory tests, though the team stresses that the full adaptation relies on a suite of metabolic and symbiotic strategies.

A separate modelling study, published in Nature Astronomy, injects caution into the search for life on Jupiter’s moon Europa. Simulations led by Rutgers University indicate that water ascending from the deep ocean through narrow ice fractures would freeze rapidly, with frazil ice blocking the conduits within hours. Any near-surface brines detected by the Europa Clipper or JUICE missions may therefore originate from local melting within the ice shell rather than from the ocean below, complicating efforts to sample the moon’s most habitable environment. The next milestones will be the analysis of TESS Cycle 2 data, the continued accumulation of FRB events for finer matter maps, and the arrival of the two Europa missions in the early 2030s.