A colossal cephalopod that stretched roughly 19 metres from arm-tip to arm-tip glided through the late Cretaceous oceans as a fearsome apex predator, according to landmark research published in Science. For the first time, palaeontologists have reconstructed the jaws of two extinct octopus relatives from rocks dating back 100 million years, revealing beaks robust enough to crush the shells of ammonites and the bones of smaller marine vertebrates. The study, led by Hokkaido University in Japan, upends the long-held assumption that only toothy reptiles and sharks occupied the top rungs of the prehistoric marine food web.

Viewed from Tokyo, the breakthrough rests on a decade of painstaking work. Because the soft bodies of octopuses almost never fossilise, their evolutionary history has remained stubbornly opaque. Researchers deployed high-resolution grinding tomography—shaving rock layers 0.05 millimetres at a time—and trained an artificial intelligence model to identify the hidden jaws within late Cretaceous rock samples that had been collected decades ago and largely ignored. The resulting three-dimensional reconstructions allowed the team to measure the beaks and estimate the animals’ total length, placing them comfortably in the ranks of the largest known invertebrates.

Analysts in London note that the timing overturns conventional wisdom. Until now, the earliest finned octopuses were thought to have appeared no earlier than 15 million years ago. The new fossils push the lineage of large predatory vampyropods—the group that encompasses modern octopuses and vampire squid—deep into the age of dinosaurs, where they would have competed directly with mosasaurs and plesiosaurs. Marine ecologists in Buenos Aires observe that such a shift forces a fundamental reassessment of Mesozoic energy flows: an invertebrate of this scale operating as a top predator rewrites the assumption that vertebrates alone dominated the upper tiers.

From Washington, palaeontologists not involved in the research describe the fossils as a “fearsome sight to behold.” The two distinct species, still unnamed, possessed thick, parrot-like beaks that imply they actively hunted hard-shelled prey, rather than scavenging. Their eight muscular arms and elongated bodies would have made them agile pursuers in open water, a far cry from the bottom-hugging, soft-bodied relatives that survived into the modern era.

The implications of the study stretch well beyond a single discovery. By applying similar machine-learning techniques to existing museum collections across Europe, North America and East Asia, scientists anticipate finding many more cephalopod fossils previously mistaken for glacial debris or fish remains. The hunt is now on for soft-tissue impressions that could colour in the anatomy of these real-life krakens. As researchers broaden their search through the late Cretaceous strata, the long-dominant narrative of reptilian supremacy in ancient seas looks increasingly incomplete. In its place, a more complex and layered picture of marine evolution is emerging—one in which the true identity of the kraken may finally have a place in the scientific record, not just in legend.