Guide

Anonymize a PDF for Claude — in your browser

Reports, transcripts, and exported documents often exist only as PDFs. Data Alias extracts the PDF's text layer inside your browser — the parser is loaded on demand and runs locally, so the document is never uploaded. Each line becomes a reviewable row, detected contact details are aliased in place, and you download an AI-ready text extract.

How it works, step by step

01

Upload a text-layer PDF

Drop the PDF into the workspace (a Pro format, free during the beta). The text layer is extracted in your browser; scanned, image-only PDFs are declined with a clear error because Data Alias does not do OCR — it will not pretend to read what it can't.

02

Review the in-place replacements

Detected emails, phone numbers, and IP addresses are suggested for in-place aliasing — the sentence around each value stays readable — while credential patterns like API keys default to removal. You review every suggestion before it is applied.

03

Download the .safe.txt extract for Claude

The output is an AI-ready .safe.txt extract of the document's text, not a rebuilt PDF. Paste or upload it to Claude; the alias mapping stays on your machine so you can restore real names in the answer.

Example: interview-04.pdf → interview-04.safe.txt

Why anonymize before Claude?

The consumer training default changed in 2025

In August 2025 Anthropic updated its consumer terms, presenting the data-for-training choice pre-checked in favor of sharing — consumer users had to actively opt out to keep chats out of training. Defaults like this can change again; a file that only ever contained aliases is indifferent to them.

Source: Anthropic consumer terms update

Settings are per-account, your data policy isn't

Whether a given Claude account opted out is invisible to you when a file gets shared onward or pasted by a colleague. Anonymizing before upload is the control that travels with the file rather than with the account settings.

Source: Anthropic consumer terms update

These are Anthropic's own documented behaviors as of our research (July 2026). Policies and defaults change — an anonymized copy is the control that doesn't depend on them.

What to know about pdf files

Text layer only — no OCR

Data Alias reads the embedded text layer. Scanned pages without one produce an honest error instead of a silently empty extract. If your PDF came from a scanner, run OCR elsewhere first or work from the source document.

The extract preserves reading context

Each line of extracted text becomes a row you can review. In-place aliasing keeps the surrounding sentences intact, so summaries and thematic analysis still work on the safe extract.

Common questions

Does it work with scanned PDFs?
No. Data Alias extracts the text layer only and does not perform OCR. A scanned, image-only PDF is rejected with a clear message rather than producing an empty "safe" file.
What does the output look like?
A .safe.txt file — an AI-ready extract of the document's text with detected values replaced in place. Data Alias does not reassemble a PDF, and doesn't claim to.
How are names inside the text handled?
Emails, phones, and IPs are detected by pattern and aliased in place. Person names can be detected with an optional on-device model (English and Korean) that runs in a browser worker — you start the model download explicitly, and detection results are suggestions to review — detection can miss values, which is why the review step exists.

An honest note on detection

Data Alias reduces exposure risk but cannot catch every sensitive value, and no automated detector can. That is why nothing is applied without your review: check the safe copy before you share it with Claude or anyone else.

Anonymize a PDF now

No account needed. Files stay in your browser — verify it yourself.

Related guides