A new AI model has revolutionized the dating of the Dead Sea Scrolls, revealing they may be older than previously thought and aligning some fragments with their presumed time of authorship.
While the Dead Sea Scrolls are immensely important documents to historical and biblical studies, the dating of the individual scrolls has remained elusive until international researchers developed Enoch, a new date prediction AI model, which has helped change our understanding of the materials’ age.
Traditionally, Jewish and Christian documents have been dated between the third century BCE and the second century CE. However, new research suggests that some may be significantly older, possibly even reaching back to the time of their presumed authorship. Enoch uses artificial intelligence to combine paleographic analysis with radiocarbon dating, producing the most precise age estimations to date the scrolls.
Paleography—the study of ancient handwriting—has been the primary tool for dating manuscripts. However, it is an imprecise method, lacking empirical evidence to support its assumptions. Only a few Aramaic and Hebrew documents from the fourth and fifth centuries BCE bear dated inscriptions. Likewise, a handful of dated documents from the late first and early second centuries CE exist, but the large temporal gap between them has limited the ability to compare handwriting styles across periods.
Training Enoch
That limitation is now being addressed through digital tools developed by The Hands That Wrote the Bible, a project funded by the European Research Council. The project trained a machine learning model using Bayesian ridge regression and radiocarbon-dated samples from 24 Dead Sea Scroll fragments, helping fill the gap in known dated writing from the fourth century BCE to the second century CE.
Using this data, researchers trained Enoch, their predictive model, which evolved from a prior neural network system called BiNet. BiNet had already been designed to detect patterns in handwritten ink traces. Enoch takes this further by replacing the human interpretation of letterforms with mathematical and geometric analysis of curves and shapes at the micro level. Cross-validation tests confirmed Enoch’s accuracy, dating manuscripts within a 30-year margin of error—an improvement even over radiocarbon dating alone.
Dating the Dead Sea Scrolls
With the model complete, researchers began applying it to nearly 1,000 documents that make up the Dead Sea Scrolls. Human paleographers reviewed Enoch’s first 135 predictions to validate the AI’s output. Due to the inherent subjectivity of traditional paleographic dating, Enoch’s results offer a valuable comparative tool, sometimes reinforcing prior assumptions and prompting reconsideration. This marks the first time a machine learning model has successfully generated probabilistic dating predictions directly from raw manuscript images.
Enoch’s early results offered some surprises. Several scrolls were dated earlier than previously believed. More notably, the model suggests that the Hasmonaean and Herodian script styles may have coexisted as early as the late second century BCE, contradicting the long-held belief that these styles emerged in the mid-first century BCE.
Impact of Dead Sea Scroll Dating
Shifting the dating of the Dead Sea Scrolls affects not only the interpretation of their content but also alters historians’ views of literacy, urbanization, and religious development in ancient Judea. These changes ripple outward, influencing how scholars understand the political and intellectual history of the Hellenistic and Roman eastern Mediterranean.
One of the most remarkable breakthroughs involves identifying the first physical fragments of biblical texts created during the time of their presumed authorship. The pieces 4QDanielc (4Q114) and 4QQoheleta (4Q109) were dated to around the 160s BCE, aligning with scholarly estimates for when the Book of Daniel was composed. Similarly, fragments of Ecclesiastes (Qohelet)—4Q114 and 4Q109—date to the second and third centuries BCE, supporting modern theories of Hellenistic authorship rather than the traditional attribution to King Solomon in the tenth century BCE.
With these compelling results, Enoch is bringing scholars closer to answering some of the oldest and most profound questions in religious history, while expanding our toolkit to analyze ancient texts using modern technological innovations.
The paper “Dating Ancient Manuscripts Using Radiocarbon and AI-based Writing Style Analysis” appeared on June 04, 2025, in PLOS One.
Ryan Whalen covers science and technology for The Debrief. He holds an MA in History and a Master of Library and Information Science with a certificate in Data Science. He can be contacted at ryan@thedebrief.org, and follow him on Twitter @mdntwvlf.