Anonymize training records for AI
To anonymize training records for AI, you start from the documents a training provider produces for each session: a training agreement in Word, a quote and an invoice in PDF, an attendance sheet, an evaluation spreadsheet in Excel, a CPF file (France's personal training account). Each one names the trainees, their employer, the funder, and carries identifiers that point to a course unambiguously: activity registration number, agreement number, session number. Biffer recognizes them on your computer, with no connection, replaces them with stable aliases and highlights them in black in the output file.
On this page we go through these files one by one, how to redact them, the Qualiopi and GDPR framework, and the AI uses that follow.
A training provider's files and what they contain that is sensitive
A training course produces a dozen files, many of which serve as evidence for Qualiopi, the French quality certification for training providers. Here are the ones we find at every trainer, with what each one reveals.
Training agreements and contracts (Word, PDF)
Agreement number, activity registration number, identity of the trainees, employer, funder, negotiated price.
Quotes and invoices (PDF)
Quote and invoice numbers, the client's French company number (SIRET), pricing terms, IBAN, name of the signatory.
Attendance sheets and certificates (PDF)
Session number, name and signature of each trainee, certificate number, certification result.
Evaluations and questionnaires (Excel, CSV, PDF)
Placement tests, results, immediate and follow-up feedback, verbatim comments that sometimes name a trainer.
CPF and EDOF files (PDF, Excel)
CPF file number, trainee number, date of birth, certification number, amounts funded (EDOF is the platform providers use to list CPF courses).
Reports and personalized materials (Word)
A learner's difficulties, disability, exchanges with the employer.
Text PDFs, Word, Excel, CSV and plain text go through the same process (see redact a PDF); images and scanned PDFs go through built-in optical character recognition, offline, then are redacted by blacking out the pixels, with no possible restoration.
Redacting these files in practice: formats, aliases, black highlighting
Everything happens on your computer, with no connection (see offline software).
- You drop the file with the trainer profile enabled. The local analysis highlights people, addresses, emails, dates, SIRET, activity registration, agreement and session numbers.
- You untick false positives (the title of a certification, the software being taught) and add what is missing, all occurrences at once.
- Each item becomes a stable alias (STAGIAIRE_001 for a trainee, CONVENTION_001, SESSION_003; the alias labels are in French), identical from one file to the next for the same trainee.
- In Word and PDF, redacted data is highlighted in black, an option enabled by default: you see at a glance what was removed. In Excel and CSV, columns and formulas are preserved.
- You choose whether to save the mapping table, encrypted (Argon2id and AES-256-GCM) or in plain text. With it, the real names are restored in the AI's answer; without it, the redaction is irreversible.
The queue takes every file from a session, in reviewed or automatic mode. Biffer is free for five files per month, all redaction features included, after a one-time email confirmation in the software, with no password. Beyond that, Biffer for Mac or Biffer for PC costs €10 excl. VAT (€12 incl. VAT) per computer, paid once by card via Stripe: a lifetime license for the version purchased, tied to the computer, unlimited files, each major version sold separately. Every feature is included from the free plan (see pricing).
The trainer profile adds training course identifiers to the general detections. Here is the exact list from the software:
| Field | Where you find it |
|---|---|
| Activity registration number (NDA) | Agreements, invoices, legal notices, annual activity report |
| Trainee number | Attendance sheets, certificates, exports from management software |
| Agreement number | Agreements, invoices, reminders |
| Session number | Attendance sheets, evaluations, schedules |
| Quote number, invoice number | Quotes, invoices, payment tracking |
| CPF / EDOF file | CPF files, exchanges with the Caisse des dépôts |
| Certificate number | Certificates of completion and attendance |
| Certification number | Certification files, result statements |
On top of these come the fields shared by all professions (file number, reference, contract number, employee number) and the general detections: people, emails, phone numbers, addresses, French company numbers (SIREN and SIRET), French social security number, IBAN, license plates, dates. An identifier specific to your organization can be added by keyword (see anonymization by profession).
Qualiopi, activity registration and GDPR
In France, every training provider must register its activity and obtain an activity registration number, as service-public.fr explains (in French). To access public or pooled funding, Article L6316-1 of the French Labor Code requires Qualiopi certification, whose criteria are set by Decree No. 2019-564 of 6 June 2019.
Two consequences for the provider's files. Qualiopi evidence is named: attendance sheets, evaluations and certificates are kept as they are for the auditor, and only a redacted copy goes to the AI. The provider is the controller for its trainees' data under the GDPR; with offline software, the data stays in-house and you keep control of the question of processors. Biffer prepares, you validate: the trainer reviews the preview and has the final word on what is shared.
AI use cases once the files are redacted
A redacted evaluation sheet keeps its scores and comments: STAGIAIRE_004 found the pace too fast, STAGIAIRE_009 asks for more depth. The AI draws trends from it without knowing anyone.
- Summarize the immediate and follow-up evaluations of a session for the Qualiopi review.
- Draft the session report or a reply to a complaint.
- Adapt a program from a group's placement tests.
- Prepare an outline for an agreement or a quote from existing documents.
The redacted file can be pasted into ChatGPT, Claude or Gemini; the page prepare a document for AI completes this guide. The answer, loaded back into Biffer with the mapping table, gets the real names back, still locally. Biffer is designed in Lyon by Romain Rissoan, himself a trainer and consultant and the founder of the software, at OPCADIA.
Frequently asked questions
Is a redacted attendance sheet still valid Qualiopi evidence?
It has a different role. The evidence for the auditor is the signed, named original, which you keep as it is in your session file. The redacted copy serves another purpose: giving it to an AI to count attendance, detect recurring absences or prepare a review, without exposing the trainees. Biffer always works on a copy and leaves the original file intact. If you saved the mapping table, the names can be restored in the document returned by the AI, on your computer.
How do you handle an Excel export of satisfaction surveys?
You drop the file as it is. Biffer reads each column, spots names, emails, companies, trainee and session numbers, and replaces them with stable aliases while fully preserving scores and free-text answers. Verbatim comments deserve a review in the preview: a trainee may name a trainer or a colleague in them, and the custom keyword field lets you add these names once for all occurrences. The file structure (tabs, formulas, filters) is preserved, and the output file remains usable in Excel and by the AI.
Should the provider's activity registration number be redacted?
It depends on the use. The activity registration number is public and says nothing about the trainees. It does, however, tie a document to your organization, and so to a session, a date and a place, which weakens the redaction of the people if the file is sent to a third party. For a file sent to an online AI, we advise redacting it; the trainer profile recognizes it out of the box (NDA_001). For a document shared with an identified funder, the "keep" policy applies to that data type alone, while the trainees stay redacted.
Your business documents anonymized, ready to be published or handed to an AI tool.
Biffer runs on your computer without sending a single file. Free for five files a month, unlimited for €10 excl. VAT per computer.
100% offline · Free: 5 files a month · No file sent