Can your security system monitor your employee’s generative AI prompts?
Generative AI platforms such as ChatGPT, Gemini, Copilot, and Claude are becoming increasingly prevalent in organisations. While these solutions enhance efficiency across various tasks, they also introduce new challenges for data leak prevention in the context of generative AI. Sensitive information can be inadvertently shared through chat prompts, files uploaded for AI-driven summarisation, or browser plugins that circumvent established security controls. Traditional Data Loss Prevention (DLP) products often struggle to detect these events. Solutions like Fidelis Network® Detection and Response (NDR) offer a network-based approach to data loss prevention, enabling organisations to monitor, enforce policies, and audit the use of generative AI as part of a comprehensive data loss prevention strategy.
The evolution of data loss prevention for generative AI necessitates a shift in focus from endpoints and isolated channels to a holistic view of the entire traffic path. Unlike earlier tools that primarily scanned emails or storage shares, NDR technologies such as Fidelis can identify threats as they traverse the network, analysing traffic patterns even when the content is encrypted. The primary concern is not solely about who created the data, but rather when and how it exits the organisation’s control, whether through direct uploads, conversational queries, or integrated AI features within business systems. To effectively monitor generative AI usage, organisations can implement GenAI DLP solutions based on network detection, employing strategies such as URL-based indicators and real-time alerts, as well as metadata-only monitoring for audit and low-noise environments.
