October 21, 2026
Can an accountant put client data into ChatGPT?
What professional secrecy and the GDPR mean for public AI tools, and the questions to ask before client data goes anywhere.
In most cases, not without careful checks. Pasting a client's figures, payslips or correspondence into a public AI tool hands that data to an outside company, often outside the EU. For an accountant, that touches two sets of rules at once: professional secrecy and the GDPR.
Professional secrecy
Accountants and tax advisers in every European country are bound by a duty of confidentiality to their clients. Sharing client information with a third party, including a software provider, needs a sound basis: the client's consent, a contract that binds the provider to the same confidentiality, or both. Your professional body's guidance is the reference for your country.
The GDPR
Client data usually includes personal data. Sending it to an AI provider makes that provider a processor, which needs a processing agreement, an entry in your register, and, if the data leaves the EU, a lawful basis for the transfer. Many public AI tools also keep conversations, and some use them to improve their models unless you switch that off.
Questions to ask before using any AI tool on client work
- Where is the data processed, and in which country?
- Is it stored afterwards, for how long, and is it used to train models?
- Is there a processing agreement, and does it cover professional secrecy?
- Can you show afterwards what the AI did with which data, and who approved the result?
The alternative: AI that stays in the office
These questions disappear when the AI runs on a machine in your own office. Client data is processed there and never leaves the building, there is no outside processor, and every action is recorded with a name and a time. That is what Bertha does.
This article is general information, not legal advice. Check your own obligations with your professional body or adviser.