Using AI on client work is not the problem. Using it without rules is. Most freelancers who use AI never decided what can be shared with a tool, who checks the output, or what happens if something goes wrong. A short written policy fixes that, and it takes less than an hour.
This article is a practical starting point, not legal advice. Your contracts, your clients’ requirements and the law where you and your client work all matter, so check them for your own situation.
Three questions to ask about any AI tool
Before you put client material into a tool, find out:
- What does it store? Does it keep your inputs, for how long, and is that explained in plain terms?
- Who can access it? The vendor’s staff, other users, subcontractors, or only you?
- How can you delete it? Can you remove your data, and is there a setting that stops your inputs from being used to improve the service?
If you cannot answer all three from the tool’s own documentation, treat the tool as unsuitable for sensitive client material.
Sort your material into three levels
A simple sorting rule is easier to follow than a long document:
- Safe: public information, your own notes, generic questions with no client details.
- Careful: client work that is not confidential but is specific. Remove names and identifying details first.
- Do not share: anything covered by an NDA, personal data about the client’s customers, credentials, financial details, unreleased plans.
When in doubt, move the item one level up.
Keep a human review step
AI can draft and analyze. Responsibility for the final result stays with you. Decide in advance:
- what you always check yourself (facts, numbers, names, claims),
- who sends the final message to the client,
- when a human takes over completely.
A deliberate review step is also your best protection against confident-sounding mistakes.
Check your contracts
Some clients restrict or prohibit the use of AI tools, or require disclosure. Read the confidentiality and data-processing clauses in your agreements. If AI use is allowed, a short line in your proposal such as “I use AI tools for drafting and analysis; client-confidential material is not entered into them” can build trust instead of raising questions.
Write the policy in ten lines
Put it in one page you can show a client on request:
- Which tools I use and why.
- What I never enter into them.
- How I remove identifying details.
- How I review output before delivery.
- How I store and delete files.
- Who is responsible for the final result: me.
You can send this page during onboarding. It costs you nothing and answers the question before it is asked.
Where does your own risk sit?
AI trust is one of six areas in the free Libebox Freelancer Business Audit. It takes about two minutes and shows where your business is weakest first.