Cloud data security company Sentra announced that large language models (LLMs) are now included in its data classification engine, enabling enterprises to accurately identify and understand sensitive unstructured data such as employee contracts, source code and user generated content. With LLMs now built directly into Sentra’s data security platform and classification engine, enterprises have the technology required to proactively reduce the data attack surface.

“Sentra is committed to innovating and paving the way for strong cloud data security in order to diminish data risks,” said Yoav Regev, co-founder and CEO at Sentra. “By taking a laser-focused approach to cloud data security, Sentra gives enterprises confidence when classifying large volumes of sensitive enterprise data at scale. With the addition of LLM technology, security teams can more accurately detect sensitive information, enabling them to root out data risks wherever they exist.”

With the increasing number of regulation and privacy frameworks, leveraging LLMs allows Sentra to automatically understand proprietary customer data with additional context like data sovereignty and region, how the data will be used, and how it should be protected. For example, a company can create data security policies that ensure employee agreements are only accessed by HR or that legal contracts are stored within a legal department’s SharePoint site. Ensuring the highest level of security, Sentra only scans data with LLM-based classifiers within the cloud premises of the enterprise.

Once a comprehensive data catalog is in place, Sentra’s ability to provide prioritized risk scoring takes multiple data layers into account, including data access permissions, activity, sensitivity, movement and misconfigurations. This gives enterprises greater visibility and control over their data risk management processes.

Key developments of Sentra’s classification engine include LLM-powered scanning of data asset content and analysis of metadata, like file names, schemas, and tags.