Scan PDFs before using them with AI.

A PDF can contain hidden text, instructions and active features beyond the visible page. Check incoming versions before they become AI context, follow the decision and retrieve the current Approved copy. PDF files up to 4 MiB are supported, subject to supported structure and complexity limits.

Coming soon

Outside exact versionOutside PDF
Organise the recordChoose a Collection for the workflow
Completed Standard ScanCheck and route

Three destinations

ApprovedLibrary → current eligible version → AI project
Needs reviewInvestigate, replace or recheck
QuarantineKeep this exact version out

The Collection keeps all three outcomes organised. Membership does not grant approval or retrieval.

No decision: The scan could not complete, so no security decision was made. Retry is available and no new scan credit is consumed.

The PDF problem

The visible page is only one layer.

PDFs are containers for text, layout, links, metadata and other machine-readable structures. Content that is difficult for a person to notice may still be extracted when the file is used with an AI tool.

The useful question is not only “Does this PDF look normal?” It is also “What might the AI read, and should any of it be reviewed first?”

Understand prompt injection in documents

What deserves attention

Review the document as a file, not only as a page.

Review supported document surfaces together before AI use.

Instruction-like content

Text that appears to direct an AI, change its task or influence how it handles the project.

Hidden and machine-readable layers

Supported content that may be concealed, encoded or otherwise easy to miss during ordinary visual review.

Links and structural features

Supported remote links, active features and PDF conditions that may require a closer look.

Your document workflow

PDF supported · Up to 4 MiB

Check the PDF before it enters your AI tool.

Keep the boundary visible whenever a PDF moves from an outside source into AI-assisted work.

1Keep the PDF and a record of where it came from.
2Choose the Collection for the client or project, then submit the exact file for checking before AI use.
3Follow Approved, Needs review or Quarantine, and read the coverage and finding locations.
4Keep the record organised in its Collection while you resolve review findings; membership does not grant approval or retrieval.
5Retrieve only the current eligible Approved exact version from the Library and access its scan record separately.
6Submit a revised or re-exported PDF as a replacement that needs its own scan.

When a replacement is accepted, it becomes current immediately. The earlier version remains in history but is not available as a fallback while the replacement is pending or held. A scan supports a human decision; it does not prove that a document is harmless or replace normal source, privacy and security checks.

Before AI use

Make the intake decision explicit.

A short checklist helps stop “looks fine” from becoming the only document-security control.

Confirm where the PDF came from and why the project needs it.
Scan the file before it enters the AI tool or project.
Read every supported finding in its document context.
Check which scans completed and where coverage was limited.
Keep the original unchanged while deciding what to use.

This checkpoint is limited to the supported checks and coverage reported for the exact version.

CoverageExplore checks and coverage ProcessSee the document review flow PDF field guideVisible pages and machine-readable PDF layers

Before documents become AI context

Give every outside PDF a checkpoint before AI use.

Check an incoming PDF, follow its decision and retrieve the current Approved copy for your AI workflow. PDF files up to 4 MiB are supported today.

Coming soonSee how it works