Is pasting a client contract into ChatGPT a breach of Swiss professional secrecy?

It can be. Art. 321 of the Swiss Criminal Code makes unauthorized disclosure of entrusted client information a criminal offence for lawyers, notaries, auditors and their auxiliaries, and Art. 13 BGFA adds the attorney's professional secrecy duty enforced by the cantonal bar supervisory authority. An LLM provider is a third party, so a prompt containing client-identifying information is a potential unauthorized disclosure. NativeAI Guard replaces parties and client data with placeholders such as [PERSON] and [CONTRACT] before the prompt leaves the firm, so lawyers can still use AI for drafting and review.

Does NativeAI become an auxiliary under Art. 321 CP?

Yes, as soon as client data passes through NativeAI's servers for inspection. The data processing agreement and the non-disclosure agreement are aligned so that support staff sit inside the same scope of professional secrecy. Firms that want to avoid this path entirely run NativeAI Guard on-premises, so no prompt and no log leaves the firm.

Where is client data processed when NativeAI Guard inspects a prompt?

Inspection runs on NativeAI's own servers in a Swiss datacenter under Swiss law, or on-premises inside the firm. Prompt content is processed transiently in volatile memory and never stored, so no copy of client data is retained by NativeAI.

Industries / Legal · law firms, notaries, audit firms & legal departments

Professional secrecy is criminal law. Protect your firm.

Art. 321 CP makes disclosing entrusted client secrets a crime for whoever sends the prompt. A prompt with client data is a disclosure unless technical controls prevent it.

See it in action

Omnichannel protection, across all devices.

ChatGPT, Claude, Gemini, Copilot, DeepL and whatever your employees try next week: the same policies, anonymization and audit trail on the work laptop and the work phone.

See scenarios →

DEADLINESNOW · Art. 321 CP · Art. 13 BGFA in forceSEP 2023 · revised DSG, personal liability up to CHF 250,000AUG 2026 · EU AI ACT HIGH-RISK (AI in the administration of justice)

The AI governance framework for law firms and legal departments

Five obligations. Here's our part, honestly.

We mark what NativeAI Guard solves for the channels we protect, where it contributes, and what stays with your processes.

Art. 321 CPCriminal professional secrecy for lawyers, notaries and auditors, including their auxiliaries and IT.✓ We solve · block/anonymize client data

Art. 13 BGFAAttorney-client privilege, unlimited in time and towards everyone, enforced by the cantonal bar supervisory authority.✓ We solve · DLP at the prompt layer

Swiss DSG / GDPRData minimization, records of processing, cross-border limits, personal liability up to CHF 250,000.✓ We solve · anonymization + RoPA

EU AI ActAI in the administration of justice is high-risk from Aug 2026, with logging, oversight and governance duties for providers and deployers.◐ We contribute · logging + oversight

Art. 730b COConfidentiality duty of the statutory auditor, for audit firms on top of Art. 321 CP.✓ We solve · DLP + anonymization

Typical scenarios in law firms and legal departments

The contract draft

A lawyer pastes a client contract into ChatGPT to check clauses. Parties, terms and the amount in dispute leave the firm in a single prompt, to a provider that knows no professional secrecy.

NativeAI Guard: parties and client data become [PERSON], [ORGANIZATION], [CONTRACT] before the prompt leaves the firm, the clause review still works.

The uploaded dossier

A paralegal uploads a statement of claim or a due-diligence dossier as a PDF to an LLM for summarization. The whole document is uploaded, with no client consent covering this processing.

NativeAI Guard: documents are inspected inside the upload, client data detected, the file blocked or anonymized before it is sent.

The research agent

An internal agent searches files, rulings and opposing submissions to prepare a brief. An opposing document with hidden instructions can redirect the agent, and its answer can expose client data from another matter.

NativeAI Guard: injected instructions are stripped before the agent sees them, and every answer is checked for client data before it comes back.

Policies, managed

The rulebook for law firms and legal departments is already written. You adapt it.

Swiss DSG, GDPR and EU AI Act, plus your industry's specific duties. Security and compliance practitioners review it; we keep it current as the law changes.

Legal packArt. 321 CP · privilegeArt. 13 BGFAClient data · partiesSwiss DSG · GDPREU AI Act, justice AIArt. 730b OR

During onboarding we tune the pack to your own data structures and internal rules, working from the compliance documents you already have.

Common questions

Frequently asked questions

Do you become an auxiliary person under Art. 321 CP, and what does that mean contractually?

Yes, as soon as client data passes through our servers for inspection, and we treat it that way. The secrecy duties that apply during processing must also apply to support and maintenance work. We align the data processing agreement and the non-disclosure agreement so that support staff sit inside the same scope of professional secrecy. If you want to avoid that path entirely, run NativeAI Guard on-premises: then no prompt and no log leaves the firm, not even to us.

Our clients require that files never leave the firm. What happens to the content during inspection?

It is processed transiently in volatile memory on our own servers in Switzerland and discarded immediately afterwards. Nothing is written to disk, nothing is used for model training. What remains is the audit entry: who sent what to which model, when, and what NativeAI Guard did about it. That is written into the data processing agreement, where you can check it, not only on this page.

We are a firm of forty lawyers, not a corporation. Is it worth it, and how do we start?

Yes. Smaller organizations get a flat rate instead of a per-user price, and the entry is the same as for a bank: a one-month passive pilot phase that we run end to end, with no change to your workflows. The result is a risk overview of data leaks, shadow AI and AI usage for the partners' meeting. From project start to live operation takes about two weeks, and your effort stays at admin access and a user list.

Protect client privilege when lawyers use AI.

A pilot phase, run by us: a policy starter pack plus a partners'-meeting report on data leaks, shadow AI, usage and costs. Client data is processed transiently, never stored.