The 19-Metre Kraken That Rewrote the Cretaceous Food Web

The Kraken That Wasn’t Just Myth

In the shallow seas off present-day Japan and Canada, 86 million to 72 million years ago, a creature grew to nearly 19 metres—longer than a school bus, with a beak 1.5 times the size of any living giant squid. It could crunch through shark armour, kraken-scale prehistoric fish, and the bones of mosasaurs. This was Nanaimoteuthis haggarti, a cephalopod that rewrites what we thought we knew about the Cretaceous ocean.

The discovery, published in Science on April 23, 2026, does not merely add another monster to the textbook. It forces a fundamental shift: the top of the Cretaceous food web was not exclusively vertebrate. An invertebrate—an octopus-like squid—held that position for tens of millions of years.

The type specimen, designated JPV-047, was unearthed in 2021 from the Late Cretaceous Horobetsu Formation along Hokkaido’s western coast, preserved in fine-grained laminated mudstone that captured minute anatomical detail. A second exceptional specimen, CNC-882, came from the Cowichan Formation on Vancouver Island’s eastern shore, embedded in a nodule of phosphatised concretion. Together, the 15 beak fossils examined by the research team spanned a geographic range that suggests N. haggarti was not a localised curiosity but a broadly distributed apex predator across the northwestern Pacific and adjacent shelf seas.

Why a Beak Tells a Story

Squid and octopus bodies are mostly soft tissue; they fossilise poorly. But they have one hard part that survives: the chitinous beak, structurally similar to the exoskeleton of a crab. In living cephalopods, beak size correlates tightly with overall body length. That relationship allows palaeontologists to estimate the size of extinct species from a single jaw fragment.

Hirohisa Motomiya, a palaeontologist at Hokkaido University and lead author of the study, led a team that examined 15 well-preserved beak fossils from late-Cretaceous deposits in Japan and Vancouver Island. By analysing the jaw structure, muscle attachment points, and breakage patterns, they reconstructed N. haggarti’s bite force and estimated its maximum length at around 19 metres. No other invertebrate fossil record comes close to that size in the same time slice.

The most diagnostic feature was the hook shape of the upper beak’s rostrum, which extended further downward than in any known living or fossil colteuthid squid. That morphology implies a hook-and-pull feeding strategy—grabbing prey and yanking it apart rather than slicing. Microscopic wear patterns on the inner surface of the lower beak showed linear striations consistent with repeated contact with cartilage and bone, not just soft fish flesh. Radiometric dating of the surrounding volcanic ash layers placed the oldest specimens at approximately 84 million years ago, meaning N. haggarti persisted through a period of significant oceanic anoxic events and shifting sea temperatures.

The AI-Powered Dig

The research also introduced a method that may change how future fossils are found. The team used AI-assisted “digital fossil mining”—processing thousands of scanned sediment samples from Japanese formations to detect hidden beak fragments buried in rock. Traditional sieving and hand-picking missed many of these small, camouflaged specimens. The algorithm identified 12 additional beak fossils, including the largest ones, which were then 3D-scanned and digitally restored.

The system, trained on a reference library of over 4,000 known cephalopod beak morphologies, recognised subtle density contrasts in X-ray computed tomography scans that the human eye overlooked. Conventional sieving had recovered only three of the 15 specimens used in the final analysis. The AI pipeline, described in supplementary materials to the study, operates at a resolution capable of distinguishing beak fragments as small as 12 millimetres from matrix noise—a threshold that would have required months of manual labour to match.

“This approach allowed us to recover jaw fossils that would have been difficult to find with conventional techniques,” Motomiya said. The technique is now being applied to other cephalopod and soft-body fossil sites across Asia and North America, potentially unlocking a wealth of previously invisible data.

What makes the method particularly significant is its reproducibility. Unlike traditional field collecting, which depends on the experience and luck of individual researchers, the AI protocol can be standardised across institutions. Several laboratories in South America and Europe have already begun applying the workflow to Cretaceous and Jurassic deposits, with preliminary results suggesting that other under-described giant cephalopod lineages may be hiding in museum collections and new quarry sites alike.

Who Wins, Who Loses

For cephalopod evolution, the finding is a triumph. It shows that the lineage of large-bodied, predatory squids and octopuses had already reached ecological dominance by the mid-Cretaceous—long before the dinosaurs went extinct. These creatures were not marginal scavengers but apex predators that shared the sea with the era’s most fearsome vertebrates: sharks with protruding teeth, mosasaurs with bone-crushing jaws, and giant plesiosaurs.

The study includes stable isotope analysis of carbon and nitrogen from three beak specimens, yielding values that place N. haggarti firmly at the top of its trophic pyramid. Its delta N-15 signature sits approximately four per mil above that of co-occurring lamniform shark teeth from the same formations—consistent with a predator that consumed other large predators, not merely smaller prey items. In practical terms, this means the squid was hunting mosasaurs, large bony fish, and possibly juvenile plesiosaurs, not just scavenging carcasses or targeting small school fish.

For textbooks, the traditional narrative—that the Mesozoic marine realm was a vertebrate playground—needs revision. The new data suggests a more complex hierarchy, where invertebrates could occupy the top trophic level, at least in certain ecosystems and periods.

For modern giant squid researchers, the study offers a rare snapshot of how large benthic cephalopods might have behaved. Their beaks, armed with the strength to puncture cartilage and crack bone, indicate a hunting strategy far more aggressive than the ambush-feeding model often proposed for today’s Architeuthis. If modern giant squids retain even a fraction of this predatory ferocity—currently understudied precisely because they are so elusive—the implications for deep-sea ecology could be substantial.

The Second-Order Ripple

The implications extend beyond taxonomy and trophic dynamics. If giant cephalopods routinely reached 19 metres and occupied apex niches in the Cretaceous, then their role in nutrient cycling, population control of mid-tier predators, and possibly even carbon sequestration through deep-ocean carcass falls deserves重新evaluation. The sheer biomass of a guild of animals this size would have represented a significant energy sink in pelagic and benthic ecosystems, one that previous models of Mesozoic marine productivity simply did not account for.

There is also a methodological ripple. The AI-mining approach demonstrated here could accelerate fossil discovery across multiple soft-bodied taxa—not just cephalopods, but also anomalocaridids, early arthropods, and other groups whose diagnostic hard parts are small and easily overlooked. Several research groups have already begun cross-applying the pipeline to Silurian and Devonian deposits, where similar identification gaps exist.

The Bigger Picture

The Cretaceous was a time of extreme climatic fluctuations, high sea levels, and rapid diversification of marine life. Into that turbulence stepped a group of animals that had abandoned their shells hundreds of millions of years earlier, yet managed to grow to sizes that rival any vertebrate predator. The discovery of N. haggarti underscores that evolution does not always move toward larger skeletons; sometimes it moves toward more flexible, adaptable forms that can exploit niches others cannot.

It also reminds us that the fossil record is far from complete. Most of the ancient ocean’s inhabitants left no trace. But thanks to digital mining and careful beak analysis, we are beginning to see the shadows of creatures that once ruled the deep.

What Comes Next

Researchers will now look for other giant cephalopod fossils in Cretaceous deposits worldwide, using the same AI-enhanced techniques. If N. haggarti is representative, the Cretaceous oceans may have hosted a whole guild of massive, active predators that left few bones behind. That would mean our understanding of prehistoric marine ecosystems is still sketchy, and the kraken may yet prove to be more common than we thought.

The immediate priority is to apply the isotope and wear-analysis protocols to additional specimens as they emerge from the AI pipeline, particularly from underexplored formations in the western Interior Seaway and the Tethyan margin. Motomiya’s team is also collaborating with marine biologists to model the biomechanics of N. haggarti’s strike sequence, drawing on high-speed videography of extant squid predation as a proxy.

For now, the 19-metre squid stands as a stark counterexample to the vertebrate-centric view of deep time. It shows that in the Cretaceous sea, the monster under the boat was not a myth—it was a real, documented, and terrifyingly large invertebrate that ruled the waves long before humans dreamed of leviathans. And as the AI tools that revealed it continue to scan deeper into the sediment record, the question is no longer whether more giants are waiting to be found, but how many Ecocene ecosystems we have misread simply because we were looking for bones when the rulers of those seas left almost none behind.