Technically, pseudonymization: aliases are reversible with the mapping file that only you hold — that reversibility is what makes the AI round-trip possible. Under most privacy laws (GDPR, Korea's PIPA), pseudonymized data is still personal data, so handle safe copies accordingly. Remove and mask are one-way; a copy is only anonymous if nothing in it can be linked back.

No. Parsing, detection, and transformation all run in JavaScript inside your open tab. Nothing is uploaded to a Data Alias server.

CSV, TXT, XLSX, JSON, PDF, and DOCX files up to 10 MB (PDF is text-layer only — no OCR yet; PDF/DOCX download as an AI-ready .safe.txt extract). XLSX, JSON, PDF, and DOCX are Pro features — free to use during the beta.

No. The Free plan works without any sign-up. Accounts arrive together with Pro — and even then, file processing stays in your browser on every plan.

Pro will be $9 a month ($79 a year) once the beta ends — see the Pricing page. Until then everything, including every Pro feature, is free.

When paid plans launch, your first purchase comes with a 14-day no-questions-asked refund. The full policy is in the terms.

No, and you should distrust any tool that claims it can. Data Alias reduces exposure risk, but no automated detector catches every sensitive value. That is why the review step exists: check the output before you share it.

Yes. The same source value always maps to the same alias within a single session, so joins and repeated identifiers remain analyzable.

Yes. Rules are applied only when you generate the safe copy. You can toggle any column off, change the rule, or reset the session at any time.

No. File contents, column values, and aliases are never logged, and nothing about your file is stored between sessions.