Data loss prevention,
rebuilt for AI

AI DLP (AI data loss prevention) is DLP built for the way employees actually use AI: typing prompts and uploading files into tools like ChatGPT, Microsoft Copilot, Gemini, and Claude. It inspects the content of each prompt and attachment in real time, at the browser and endpoint layer, and blocks or redacts sensitive data such as PII, financial records, and source code before it reaches the AI model.

Legacy DLP was designed for email, file shares, and endpoints moving files around, and it typically inspects network traffic or data at rest. AI traffic is HTTPS encrypted and happens inside the browser, so network-layer DLP cannot read the prompt. AI DLP operates at the prompt level instead: it inspects the text and files an employee submits to an AI tool before they leave the device, enforces policy in real time, and can redact just the sensitive portion rather than blocking the whole interaction.

More than the big four. ChatGPT, Microsoft Copilot, Gemini, and Claude account for most enterprise AI usage, but shadow AI means employees also reach for thousands of niche tools that never went through procurement. Aona applies one DLP policy across 5,600+ AI tools, so coverage does not depend on maintaining a manual blocklist.

No. Blanket blocking pushes AI usage onto personal devices where you have no visibility. AI DLP reserves hard blocks for genuinely risky tools and data, and uses redaction and real-time coaching to give employees a safe path for everything else. A regulated Australian healthcare college using this approach cut shadow AI prompts by 92.9% in three months while overall AI usage kept growing.

No. AI DLP complements legacy DLP rather than replacing it. Your existing DLP keeps covering email, file shares, and endpoints; AI DLP adds the layer legacy tools cannot see, which is the prompt and file upload inside AI tools. Most organisations run both side by side, with Aona enforcing the AI layer.