Document anonymization software for in-house counsel and DPOsAI on your contracts, without exposing your counterparties or your employees.

AnonymFiles, document anonymization software for in-house counsel and DPOs, installed on your computer. It stands out on three points: it processes the whole case file with the same labels across every document, it redacts Word and PDF files inside the document itself, layout preserved, and it puts the real names back into the AI's answer. Everything happens on your computer, with no account and no network call for the license. Contracts, NDAs, HR files and .eml or .msg emails are processed in one go. €290 excl. VAT per seat per year, €240 excl. VAT per seat from 3 seats, no page quota. Anonymize a contract before ChatGPT · Our anonymization software comparison.

No: it happens on your computer, with no account and nothing sent, and the publisher receives no documents. The AI you then give the copy to falls under your usual assessment (article 28 contract, terms of use): a pseudonymized copy reduces what reaches it, without taking the data out of the GDPR. Anonymization vs pseudonymization: the difference.

Yes, if an indirect detail remains visible: a project or product name, a unique job title, a site address, a very recognizable contract amount. That's why each document is reviewed side by side before export, and one click masks whatever detection missed.

With caution. French law no. 2026-122 of 23 February 2026 makes in-house counsel's opinions confidential, under conditions: a master's degree in law or equivalent, training in ethics rules, and the mention “confidentiel - consultation juridique - juriste d'entreprise”. It applies from a date set by decree, at the latest on 1 February 2027, and cannot be invoked in criminal or tax proceedings. It says nothing about AI tools: give an online AI a copy in which the company, its counterparties and its employees are no longer named.

The one given by article 39 of the GDPR: to “inform and advise” the controller and employees, and to monitor compliance with the regulation. A simple rule to write into the internal policy: no document goes to an AI without a pseudonymized, reviewed copy. In AnonymFiles, nothing is exported without document-by-document approval, and the export index lists exported and excluded documents.

No. Labels prevent the information from being linked to the company, not the information itself from being read: a process, a price or a strategy remain readable if they are in the copy. Protection requires “reasonable protective measures” (French Commercial Code, art. L. 151-1). Select during review the passages the AI doesn't need to see: they become a label throughout the file, just like a name.

It needs particular care: the GDPR classes data concerning health and trade union membership as special categories, whose processing is prohibited save for exceptions (art. 9). The employee, their colleagues and their contact details become labels; an unfitness for work, a sick leave or a union mandate may still be enough to recognize them. Detection isn't designed for these mentions: select them in the original. If the task doesn't require these documents, don't include them.

Yes, on the data room documents exported as a folder: contracts, minutes, tables and e-mails are processed in one go, and each company keeps the same label everywhere. First reread the confidentiality agreement signed for the transaction: it is that agreement, not the tool, that says what may be given to a service provider. Amounts and project names are masked with one click during review.