eSentire Labs Open Sources Project to Monitor LLMs

eSentire Labs has announced the eSentire LLM Gateway, an open source implementation framework that provides security teams with visibility and control to monitor the use of ChatGPT and other large language models (LLMs) in their environments. The LLM Gateway is available on GitHub.

The LLM Gateway was originally developed for internal eSentire use to govern and monitor corporate usage of ChatGPT, Google Bard, and other LLMs, the company said. Security operations center and threat response teams have access to telemetry that can be used to identify LLM-based attacks that are more sophisticated than phishing and impersonation attacks, said Alexander Feick, vice-president of eSentire Labs, in a statement.

The open source project gives IT administrators the ability to integrate and apply their own security controls, including corporate policies, usage rules, and prompts.

The LLM Gateway creates a protective layer between corporate data and AI applications and logs all interaction with LLMs, according to eSentire. All interactions are logged to enhance monitoring and security controls. Plug-ins provide IT administrators with recommendations on how to visualize and track LLM usage.

The company plans to introduce a complete managed detection and response for its LLM service offering, Feick said.