AI hunts Hubble archives, finds 1,300 cosmic anomalies, many still unclassified

AI researchers used AnomalyMatch to scan 100 million Hubble Legacy Archive image cutouts, revealing about 1,300 unusual objects—800 newly discovered—including merging galaxies, stars with trailing gas, and several gravitational lenses; dozens defy current classification. The work, published in Astronomy & Astrophysics, demonstrates AI can accelerate discovery in vast archival datasets and may help future surveys from Euclid, the Nancy Grace Roman Telescope, and the Rubin Observatory manage the data deluge.