HiddenContent.ai

HiddenContent.ai builds document inspection infrastructure for organisations running AI over untrusted files. The service identifies text that is present in a document but invisible when rendered, a class of content that ordinary extraction tools pass through silently and that language models read without hesitation.

The technical approach is a rendering diff rather than pattern matching. The pipeline parses the format natively, resolves style inheritance to determine visibility, renders the page, and compares the result against the underlying file. Concealment techniques are identified individually and drawn from a reference generated from the running build.

Every result includes sanitised text, a cleaned file, and a signed certificate that can be verified offline. No document is transmitted to a third party and no model inference is performed. Deployment options cover a hosted API and a self-hosted build for air-gapped networks. Pricing is prepaid at one cent per page.