Can I use AI for client work?

Using AI to do client work faster is common and often reasonable. What separates careful use from a problem is a short check before you start.

The five checks

1. Your contract with the client. Does it restrict sharing information with third parties, subcontractors or specific technologies? Many confidentiality clauses treat an AI provider as a third party. Some clients now include AI clauses of their own.

2. Your employer’s policy. If you work for a firm or company, its rules come first. See whether your company needs an AI policy.

3. Your profession’s rules. Lawyers, accountants, health professionals, financial advisers and others may have duties of confidentiality, competence and disclosure that specifically affect AI use. Check your regulator or professional body.

4. The tool’s data terms. Which account? Is the data used for training? How long is it kept? Business and API plans usually differ from consumer ones; see Claude for one example and company data in ChatGPT for a general checklist.

5. The sensitivity of the material. Public information and your own reasoning is low risk. Client financials, personal data and unpublished plans are not.

Ways to use AI safely on client work

• Anonymize: replace names, amounts and identifiers with placeholders.
• Work on structure: ask for a template or outline, then fill in the client’s details yourself.
• Use an approved enterprise tool for anything confidential.
• Keep AI on your side of the desk: use it to prepare, research and draft internally.
• Keep a record of how AI was used in each deliverable.

Quality and accountability

The client hired you, not the tool. That means:

1. Check facts, figures, citations and quotes; see how to stop AI hallucinations.
2. Read the whole deliverable as if someone else wrote it.
3. Be ready to explain how you reached each conclusion.
4. Avoid claiming work is “original analysis” if AI produced it and you did not verify it.

What published rules and policies say

• The ABA ethics opinion on generative AI says a lawyer using generative AI must keep confidential all information relating to the representation of a client, regardless of its source, unless the client gives informed consent. Other professions have their own rules, and checking before you paste client data applies to all of them.
• The ABA ethics opinion on generative AI lists duties to provide competent legal representation, to protect client information, to communicate with clients and to charge reasonable fees when using generative AI.
• Anthropic says that by default it will not use inputs or outputs from its commercial products, such as Claude for Work and the Anthropic API, to train its models. Consumer plans are covered separately, so use the account your client agreement allows.
• The U.S. Copyright Office concluded that generative AI outputs can be protected by copyright only where a human author has determined sufficient expressive elements.
• The U.S. Copyright Office also confirmed that using AI to assist in the process of creation does not bar copyrightability.
• The Copyright Office said a human who makes creative arrangements or modifications of AI output may qualify, but not the mere provision of prompts. Put ownership terms in your contract.
• Microsoft Copilot documentation says users should still use their judgment when reviewing the output before sending it to others.

Disclosure

There is no single rule. Some clients want to know, some engagements or regulators require it, others do not care. A short line in your engagement letter (“we may use AI tools to assist with drafting, under human review”) avoids surprises.

Contracts and legal review

If the client work involves contracts, AI can help you read and compare, but it does not replace legal advice; see whether AI can review contracts.

When not to

Do not use AI where a client or law forbids it, or where you could not detect a mistake. And never enter credentials or regulated data into a tool not approved for them.

The class covers privacy and the core elements of a workplace AI policy, so you can ask these questions before using AI on client work. It does not use your client data in exercises.

Sources

• American Bar Association, ABA issues first ethics guidance on a lawyer’s use of AI tools: “keep confidential all information relating to the representation of a client, regardless of its source, unless the client gives informed consent”
• American Bar Association, ABA issues first ethics guidance on a lawyer’s use of AI tools: “duties to provide competent legal representation, to protect client information, to communicate with clients and to charge reasonable fees”
• Anthropic Privacy Center, Is my data used for model training? (commercial products): “By default, we will not use your inputs or outputs from our commercial products”
• U.S. Copyright Office, NewsNet 1060, Copyright and Artificial Intelligence Part 2: Copyrightability: “can be protected by copyright only where a human author has determined sufficient expressive elements”
• U.S. Copyright Office, NewsNet 1060, Copyright and Artificial Intelligence Part 2: Copyrightability: “does not bar copyrightability”
• U.S. Copyright Office, NewsNet 1060, Copyright and Artificial Intelligence Part 2: Copyrightability: “a human makes creative arrangements or modifications of the output, but not the mere provision of prompts”
• Microsoft Learn, Data, Privacy, and Security for Microsoft Copilot: “users should still use their judgment when reviewing the output before sending them to others”