For accounting firms

ChatGPT for the accountant’s work, without client tax ids leaving the office.

Salary certificates, payroll, balance sheets, emails to the tax office: all carry tax ids, social security numbers, IBANs and names. filterit pseudonymises them before any AI tool and restores them in the answer.

Start free See the accuracy

One paste is a transfer

The certificate you “fix” in a free chatbot carries a tax id, a social security number and a salary. Under the GDPR that is processing by you, with the provider as recipient.

Free accounts train models

Consumer editions may use conversations for training unless you switch it off. Business editions usually do not, but they store.

The client trusts you with confidentiality

The profession’s code of conduct and the engagement letter have no clause “unless I sent it to an AI”.

What gets detected

What gets detected in your documents

Greek-specific identifiers

  • Tax ID (ΑΦΜ) — Greek tax identification number with modulo-11 check digit validation, not a bare 9-digit match. (check: modulo 11)
  • Social security (ΑΜΚΑ) — Greek social-security number with Luhn and birth-date validation. (check: Luhn)
  • Business registry (ΓΕΜΗ) — General Commercial Registry number, 12 digits, context-gated to avoid clashing with other numbers. (check: context)
  • myDATA and EFKA registries — AADE myDATA MARK and invoice UID, EFKA employer and insured numbers (ΑΜΕ, ΑΜΟΕ). (check: context)
  • Property references (ΑΤΑΚ, ΠΕΑ) — E9 property identifier and energy-performance certificate security numbers. (check: context)
  • Tiresias, auctions, BIC — Greek credit-bureau references, e-auction codes and BIC/SWIFT codes. (check: context)

European and international

  • IBAN — Bank accounts of every EU/EEA country with MOD-97 validation. (check: MOD-97)
  • EU and UK VAT numbers — VAT numbers of all 27 EU countries and the UK: per-country format check, check digit for Greece, Italy, Poland, Germany and the UK. (check: VIES)
  • National IDs (EU, US, UK) — DNI, PESEL, Codice Fiscale, NIR, SSN/ITIN/EIN, National Insurance and more. (check: national checks)
  • Phone numbers — Landlines and mobiles for the EU, UK, US/Canada and Switzerland, validated per country. (check: libphonenumber)
  • Payment cards — Card numbers with Luhn, together with CVV and expiry. (check: Luhn)
  • Digital signatures — E-signature identifiers: DocuSign envelope, Adobe Sign agreement, SHA-1/SHA-256 certificate thumbprints. (check: format)
  • Postal codes — Greek postcodes, US ZIP codes and UK postcodes. (check: context)

People and contact details

  • Names across grammatical cases — Greek NER that follows inflection, greeklish spellings and links the forms of the same person.
  • Companies and organisations — Company and institution names, Greek and international.
  • Email addresses — Email addresses. (check: format)

Code, secrets and systems

  • Reference codes — Prefixed identifiers such as EMP-, INV-, CUST-, ORD-. (check: format)

In scanned PDFs and images, signatures and stamps are covered by visual redaction, with a preview before it is applied.

Every category and the accuracy measurements

How it works, in three steps

Paste or upload

Text, Excel as text, PDF, scanned invoices through OCR. You see the findings before anything leaves.

The AI sees placeholders

“Employee [NAME_1] (tax id [AFM_1], ΑΜΚΑ [AMKA_1]) with gross pay.”. The tax id is recognised by its checksum, not because it has nine digits, so invoice numbers stay.

The answer returns complete

The email to the client or the tax office comes back with the real data, in your own account.

Email assistant

Gmail and Outlook: the thread is pseudonymised before the AI, the draft returns complete for you to review and send.

Data masking for documents

Word, PDF, scans. Preview, export, restore.

Word add-in

Mask a selection or the document inside Word, restore on the same machine.

Browser extension

For the ChatGPT, Claude and other AI chats you already use. Masks before you press Enter. Chrome and Edge, manual install.

What you answer the DPO

  • Pseudonymisation by default: the measure the GDPR names explicitly in Articles 25 and 32.
  • Minimisation: you send the paragraph, not the file. The record shows what was masked and when.
  • Mapping encrypted on EU infrastructure, per firm account.
  • A guide with the 12 points and a one-page AI-use policy template, ready for the GDPR file.
  • A control mapping for ISO/IEC 27701 and 27001 with the evidence for each control, public at filterit.ai/trust/control-mapping, for the DPO file. It supports documentation, it does not certify.

Guides for the GDPR file

GDPR checklist for AI tools in a Greek office

12 points with the articles next to each, a one-page AI-use policy template, what to mask. Greek.

Read PDF

Frequently asked

Does it work with Excel?

Paste the cells as text or export to CSV. Tax ids, social security numbers, IBANs and names are recognised column by column.

What about company names?

They are masked by default, under the "Companies and organisations" category, even though they are not personal data. An organisation can have the category switched off for its workspace on request.

Does it need installing?

Not for the app. The extension is installed manually in the browser (Chrome, Edge). The Windows desktop agent is optional, on the Team and Enterprise plans, for firms that want central enforcement.

Try it on a salary certificate

Free account, no card. See what it finds before anything leaves.

Start free

Legal references are general information, not legal advice. Accuracy figures are published on the accuracy page and refreshed with every release.